CN115358817A - Intelligent product recommendation method, device, equipment and medium based on social data - Google Patents
Intelligent product recommendation method, device, equipment and medium based on social data Download PDFInfo
- Publication number
- CN115358817A CN115358817A CN202210992127.0A CN202210992127A CN115358817A CN 115358817 A CN115358817 A CN 115358817A CN 202210992127 A CN202210992127 A CN 202210992127A CN 115358817 A CN115358817 A CN 115358817A
- Authority
- CN
- China
- Prior art keywords
- story
- social
- scene
- line
- text data
- 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
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- 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
Abstract
The application provides an intelligent product recommendation method and device based on social data, an electronic device and a computer-readable storage medium, wherein the method comprises the following steps: constructing a plurality of scene lines for identifying user requirements, wherein the scene lines comprise a plurality of story nodes, and each story node comprises a story label and a preset semantic classification model; the method comprises the steps of obtaining social text data of a user, determining the matching degree of the social text data and each scene line according to semantic classification models of a plurality of scene lines, and taking the scene lines meeting the matching requirements as first story lines; determining a first vocabulary vector of each participle in the social text data and determining a second vocabulary vector of each story label preset keyword in each first story line so as to determine the similarity between the social text data and the first story lines; and taking the first story line with the similarity meeting the preset threshold requirement as a target story line to determine a target recommended product. The method and the device can be used for mining the real requirements of the user and improving the product recommendation accuracy.
Description
Technical Field
The present application relates to the field of intelligent recommendation technologies, and in particular, to a social data based intelligent product recommendation method and apparatus, an electronic device, and a computer-readable storage medium.
Background
In the product recommendation application based on financial services at present, intelligent recommendation is an important module. In practical application, the intelligent financial product recommendation system not only needs to discover the financial appeal of the user from historical data of financial transactions, but also needs to fully discover the characteristics of the user from different data sources, so that the recommended financial products can meet the appeal of potential users to the maximum extent.
Most of the traditional financial product recommendation systems are based on the historical records of financial transactions of users, such as financial records, policy records, transaction breakpoints and other information. The advantage of using the historical transaction records to analyze the user is that the data source is direct, the transacted range of the user and key points of lack of guarantee of the user can be visually analyzed, and targeted marketing is performed in the later period according to the analysis result. The main problem of the conventional financial product recommendation system is that the user data source is single, and the real state of the user in the life cannot be obtained from historical transaction data, so that the real appeal of the user in the daily life is difficult to find, and the accuracy of the financial product recommendation is low.
Disclosure of Invention
In view of the foregoing, it is desirable to provide an intelligent product recommendation method and apparatus based on social data, an electronic device, and a computer-readable storage medium, so as to solve the technical problem that the accuracy of financial product recommendation is not high.
In a first aspect, an embodiment of the present application provides a method for recommending an intelligent product based on social data, where the method includes:
constructing a plurality of scene lines for identifying user requirements, wherein the scene lines comprise a plurality of story nodes, each story node comprises a story label and a preset semantic classification model, and each scene line corresponds to at least one recommended product;
obtaining social text data of a user, determining the matching degree of the social text data and each scene line according to the semantic classification models of the scene lines, and taking the scene lines meeting the matching requirements as first story lines;
determining a first vocabulary vector of each participle in the social text data, and determining a second vocabulary vector of each story tag preset keyword in each first story line;
determining the similarity between the social text data and the first story line according to all first vocabulary vectors corresponding to the social text data and all second vocabulary vectors corresponding to each first story line;
and taking the first story line with the similarity meeting the preset threshold requirement as a target story line, and determining a target recommended product according to the target story line.
In some embodiments, the semantic classification model comprises a two-classification model; the determining the matching degree of the social text data and each scene line according to the semantic classification models of the scene lines comprises:
inputting the social text data into each two classification models of the scene line for classification to obtain matching probability output by each two classification model and using the matching probability as matching probability of a corresponding story node;
and performing joint probability distribution operation on the matching probabilities of all story nodes of the scene line to obtain the matching probability of the scene line and the social text data, and taking the matching probability as the matching degree.
In some embodiments, the two classification models include a Bert model and an integrated classifier connected to a rear end of the Bert model, and the matching probability of the story node is an average value of classification probabilities output by all sub-classifiers in the integrated classifier.
In some embodiments, the training process of the Bert model includes:
obtaining a social text training data set, wherein samples of the social text training data set are marked with positive/negative sample categories;
inputting a social text training data set into a Bert model for coding, outputting a prediction classification result through a rear-end integrated classifier, comparing the prediction classification result with positive/negative sample categories of corresponding samples, and determining a prediction error;
and adjusting the model parameters of the Bert model according to the prediction error, performing gradient descent learning on the coding part of the Bert model, performing ensemble learning on a rear-end ensemble classifier, and performing iterative training until the prediction error is smaller than a preset error threshold value to obtain the completely-trained Bert model.
In some embodiments, each story tag is preset with at least one keyword; the determining a first vocabulary vector of each participle in the social text data and determining a second vocabulary vector of each story tag preset keyword in each first story line comprises:
performing Word segmentation on the social text data, and converting each Word segmentation into a corresponding first Word vector by using a preset Word vector model Word2 vec;
and converting all preset keywords of each story label in the first story line into corresponding second Word vectors by using a preset Word vector model Word2 vec.
In some embodiments, the determining a similarity between the social text data and the first story line according to all first vocabulary vectors corresponding to the social text data and all second vocabulary vectors corresponding to each of the first story lines comprises:
and determining the shortest distance from each first vocabulary vector of the social text data to all second vocabulary vectors of the first story line by using a cosine similarity measure method, and taking the average value of the shortest distances corresponding to all first vocabulary vectors as the similarity between the social text data and the first story line.
In some embodiments, the recommended product includes at least one service product in at least one financial topic.
In a second aspect, an embodiment of the present application provides an intelligent product recommendation device based on social data, where the device includes:
the system comprises a scene building unit, a recommendation unit and a recommendation unit, wherein the scene building unit is used for building a plurality of scene lines for identifying user requirements, each scene line comprises a plurality of story nodes, each story node comprises a story label and a preset semantic classification model, and each scene line corresponds to at least one recommended product;
the matching data unit is used for acquiring social text data of a user, determining the matching degree of the social text data and each scene line according to the semantic classification models of the scene lines, and taking the scene lines meeting the matching requirements as first story lines;
the vector determining unit is used for determining a first vocabulary vector of each participle in the social text data and determining a second vocabulary vector of each story label preset keyword in each first story line;
the similarity calculation unit is used for determining the similarity between the social text data and the first story line according to all first vocabulary vectors corresponding to the social text data and all second vocabulary vectors corresponding to each first story line;
and the product recommendation unit is used for taking the first story line with the similarity meeting the preset threshold requirement as a target story line and determining a target recommended product according to the target story line.
In a third aspect, an embodiment of the present application further provides an electronic device, where the electronic device includes:
a memory storing at least one instruction;
a processor executing instructions stored in the memory to implement the social data based intelligent product recommendation method.
The embodiment of the application also provides a computer-readable storage medium, and at least one instruction is stored in the computer-readable storage medium and executed by a processor in an electronic device to implement the social data based intelligent product recommendation method.
Summarizing, the scene lines of a plurality of positioning user demands are constructed, semantic matching is carried out on social text data of a user and each story node of each scene line, a plurality of first story lines which are matched can be screened out, word segmentation processing is carried out on the social text data, story label keywords of each word segmentation and each first story line are accurately filtered in a word vector measurement mode, a target story line is obtained, a target recommended product is determined, accuracy of product recommendation is effectively improved, and marketing efficiency of follow-up marketing services is facilitated to be improved.
Drawings
FIG. 1 is a flow chart of a preferred embodiment of a social data based intelligent product recommendation method to which the present application relates.
FIG. 2 is a functional block diagram of a preferred embodiment of the social data based intelligent product recommendation device according to the present application.
Fig. 3 is a schematic structural diagram of an electronic device according to a preferred embodiment of the social data-based intelligent product recommendation method according to the present application.
Detailed Description
For a clearer understanding of the objects, features and advantages of the present application, reference is made to the following detailed description of the present application along with the accompanying drawings and specific examples. It should be noted that the embodiments and features of the embodiments of the present application may be combined with each other without conflict. In the following description, numerous specific details are set forth to provide a thorough understanding of the present application, and the described embodiments are merely a subset of the embodiments of the present application and are not intended to be a complete embodiment.
Furthermore, the terms "first", "second" and "first" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or to implicitly indicate the number of technical features indicated. Thus, features defined as "first", "second", may explicitly or implicitly include one or more of the described features. In the description of the present application, "a plurality" means two or more unless specifically limited otherwise.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein in the description of the present application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. As used herein, the term "and/or" includes any and all combinations of one or more of the associated listed items.
The embodiment of the Application provides an intelligent product recommendation method based on social data, which can be applied to one or more electronic devices, wherein the electronic devices are devices capable of automatically performing numerical calculation and/or information processing according to preset or stored instructions, and hardware of the electronic devices includes but is not limited to a microprocessor, an Application Specific Integrated Circuit (ASIC), a Programmable Gate Array (FPGA), a Digital Signal Processor (DSP), an embedded device, and the like.
The electronic device may be any electronic product capable of performing human-computer interaction with a user, for example, a Personal computer, a tablet computer, a smart phone, a Personal Digital Assistant (PDA), a game machine, an Internet Protocol Television (IPTV), an intelligent wearable device, and the like.
The electronic device may also include a network device and/or a user device. The network device includes, but is not limited to, a single network server, a server group consisting of a plurality of network servers, or a Cloud Computing (Cloud Computing) based Cloud consisting of a large number of hosts or network servers.
The Network where the electronic device is located includes, but is not limited to, the internet, a wide area Network, a metropolitan area Network, a local area Network, a Virtual Private Network (VPN), and the like.
Fig. 1 is a flowchart illustrating a social data-based intelligent product recommendation method according to a preferred embodiment of the present invention. The order of the steps in the flow chart may be changed and some steps may be omitted according to different needs.
Step S10, a plurality of scene lines used for identifying user requirements are constructed, each scene line comprises a plurality of story nodes, each story node comprises a story label and a preset semantic classification model, and each scene line corresponds to at least one recommended product.
It can be understood that all story nodes of all scene lines can be shared, one scene line can be a Markov chain consisting of a series of preset semantic classification models, one Markov chain can accurately position a scene with user requirements, and different Markov chains, namely scene lines, can position different user requirements; in addition, each story node carries a story label with semantic information, and each story label can be expressed by expanding a plurality of preset keywords, so that the preset keywords of the story labels of each story node can form an independent keyword library for subsequent accurate matching.
In an alternative embodiment, the recommended product includes at least one service product in at least one financial topic.
The financial theme can include insurance, financing, loan, health, investment, etc., and the service product of one financial theme can include multiple categories, such as insurance theme, and the service product includes health insurance, car insurance, pet insurance, etc.
In this alternative embodiment, one scenario line may locate one recommended product, and one recommended product may also be located by a plurality of scenario lines individually.
Illustratively, for example, a scenario line for constructing a pet insurance service product in an insurance theme, scenario line [ friends house pets ]: story node 1[ friend ], story node 2[ raise ], story node 3[ pet ], wherein "friend domestic pet" is a scene for locating that the user has a potential pet risk demand, and "friend", "raise", "pet" can be keywords in different story labels, wherein "friend" can also be keywords representing human meaning information such as XX relatives, XX names, colleagues and classmates; the "raising" may also be a keyword representing semantic information related to pet interaction, such as raising, feeding, getting-home, teasing, taking, walking, etc.; the "pet" can also be a keyword representing semantic information of the pet, such as cat, dog, tibetan mastiff, and specific pet name.
Therefore, the real requirements of the user can be mined by constructing a plurality of scene lines formed by a series of story nodes, and the accuracy of recommending the social data products is effectively improved.
S11, obtaining social text data of a user, determining the matching degree of the social text data and each scene line according to the preset semantic classification models of the scene lines, and taking the scene lines meeting the matching requirements as first story lines.
It is to be understood that the social text data may be text data directly input by the user, or text data into which voice data input by the user is converted.
In an optional embodiment, the preset semantic classification model comprises a binary classification model; the determining the matching degree of the social text data and each scene line according to the preset semantic classification model of the scene lines comprises the following steps:
inputting the social text data into each two classification models of the scene line for classification to obtain matching probability output by each two classification model and using the matching probability as matching probability of a corresponding story node;
and performing joint probability distribution operation on the matching probabilities of all story nodes of the scene line to obtain the matching probability of the scene line and social text data, and taking the matching probability as the matching degree.
In this optional embodiment, the social text data is input into each two-classification model of each scene line, based on the story label of each story node, the two-classification model outputs the semantic matching probability of the social text data and the corresponding story label, and then the matching probabilities of all the story nodes of the corresponding scene line are subjected to joint probability distribution operation to obtain the matching probability of the scene line, wherein a specific operation formula is as follows:
P(sl,text)＝P(sp 1 |text)*P(sp 2 |text)*…*P(sp n |text)*P(text)；
wherein sp n For the nth story node, text is social text data, sl is a scene line, and P (text) is the prior probability of a classification model for identifying and classifying the social text data. The matching probability of each story node is P (sp) n With P (sl, text) as a scene lineThe match probability is the product of the match probabilities of all story nodes on this scene line.
Further, the matching probability of all the scene lines and the social text data is used as the matching degree, the matching degrees of all the scene lines are ranked from large to small, and the scene lines with the preset number ranked at the top are selected as the first story line. The preset number may be set according to an actual situation, and is not limited herein.
Therefore, the first story line matched with the user can be mined, so that the next accurate positioning is facilitated.
In an optional embodiment, the two classification models include a Bert model and an integrated classifier connected to a rear end of the Bert model, and the matching probability of the story node is an average value of classification probabilities output by all sub-classifiers in the integrated classifier.
The method comprises the steps of obtaining a binary encoder representation from a Transformer (Transformer bidirectional coding representation) model, learning a deep fusion feature vector of an input text sequence, and applying the learned deep fusion feature vector to a classification task of an integrated classifier, wherein the Bert (bidirectional encoder representation) model is a pre-trained language representation model.
In this alternative embodiment, given social text data, the match probability P (sp) of each story node n | text) is the classification probability of an independent Bert model output. The Bert model is a more excellent pre-training model at the present stage, and in this embodiment, based on the Bert model trained by an open chinese corpus, pre-training and fine-tuning mode training are performed in a social text training dataset. It should be noted that, because the positive and negative samples in the existing social text training data set are extremely uneven, and the number of the negative samples is much greater than that of the positive samples, in this embodiment, an integrated classifier is connected to the rear end of the Bert model, and according to the number of the sub-classifiers, a mode of equally sampling the negative samples and fully sampling the positive samples is set for each sub-classifier to train the integrated classifier, the coding part of the Bert model is trained in a gradient descent mode, and finally, the classification probability output by one Bert model is the average value of the classification probabilities output by all sub-classifiers. Thus, the device is provided withThe problem of unbalance of the positive and negative samples can be effectively solved, and the accuracy of Bert model classification is improved.
In this optional embodiment, the training process of the Bert model includes:
obtaining a social text training data set, wherein samples of the social text training data set are marked with positive/negative sample categories;
inputting a social text training data set into a Bert model for coding, outputting a prediction classification result through a rear-end integrated classifier, comparing the prediction classification result with the positive/negative sample categories of corresponding samples, and determining a prediction error;
and adjusting model parameters of the Bert model according to the prediction error, performing gradient descent learning on a coding part of the Bert model, performing ensemble learning on a rear-end ensemble classifier, and performing iterative training until the prediction error is smaller than a preset error threshold value to obtain the completely trained Bert model.
In this optional embodiment, the social text training dataset of the Bert model of each story node is an independent training dataset, the samples with corresponding story tag semantic information are marked as a positive sample category and marked as 1, and the samples without corresponding story tag semantic information are marked as a negative sample category and marked as 0. Further, each sample carrying the class label is input into a Bert model for encoding, and a prediction classification result is output through a rear-end integrated classifier, wherein the prediction classification result comprises a prediction classification and a classification probability; comparing the prediction classification result with the class label of the corresponding sample, and determining a prediction error according to a preset loss function; and adjusting the model parameters of the Bert model according to the prediction error, performing gradient descent learning on the coding part of the Bert model, performing ensemble learning on a rear-end ensemble classifier, and performing iterative training until the prediction error is smaller than a preset error threshold value to obtain the completely-trained Bert model. In this embodiment, the preset loss function may be any one of a logarithmic loss function, an absolute value loss function, a perceptual loss function, a logarithmic loss function, and a square loss function. The preset error threshold may be within 5%, for example, 2%,3%, or 4%, and may be set according to actual conditions.
In this way, a complete Bert model is trained through each story node, the matching probability of the social text data and each story node and the matching probability of the social text data and the scene line are determined, then N first story lines which are most matched with the social text data are determined through sequencing and respectively correspond to the recommended scenes of K recommended products, wherein the N value and the K value can be the same positive integer or different positive integers.
Step S12, determining a first vocabulary vector of each participle in the social text data, and determining a second vocabulary vector of each story tag preset keyword in each first story line.
In an optional embodiment, the determining a first vocabulary vector of each participle in the social text data and determining a second vocabulary vector of each story tag preset keyword in each first story line includes:
performing Word segmentation on the social text data, and converting each Word segmentation into a corresponding first Word vector by using a preset Word vector model Word2 vec;
and converting all preset keywords of each story label in the first story line into corresponding second vocabulary vectors by using a preset Word vector model Word2 vec.
The Word vector model Word2vec is a tool for converting vocabulary into a vector form through calculation, in addition, the Word2vec can simplify the processing of text contents into vector operation in a vector space through training, and the similarity in the vector space can be used for expressing the similarity in text semantics. Thus, word2vec output vocabulary vectors can be used to do work such as clustering, synonym finding, part-of-speech analysis, and so on, and word2vec is very efficient.
In this optional embodiment, a dictionary-based Word segmentation algorithm or a statistical-based machine learning algorithm may be adopted to perform Word segmentation processing on social text data, and then each Word segmentation is converted into a corresponding first vocabulary vector by using a preset Word vector model Word2 vec; and converting all preset keywords of each story label in the first story line into corresponding second Word vectors by using a preset Word vector model Word2 vec.
Therefore, subsequent semantic matching is facilitated, when the participles of the social text data are not identical to the keywords in the keyword library corresponding to the story label, but the keywords are similar, matching can still be performed, and the success rate of text matching is improved. In addition, compared with the traditional keyword library, the keyword library of each story label in the embodiment can perform semantic matching by only storing a small number of representative keywords, so that the use amount of the keywords is reduced.
Step S13, according to all first vocabulary vectors corresponding to the social text data and all second vocabulary vectors corresponding to each first story line, determining the similarity between the social text data and the first story line.
In an optional embodiment, the determining the similarity between the social text data and the first story line according to all first vocabulary vectors corresponding to the social text data and all second vocabulary vectors corresponding to each of the first story lines includes:
and determining the shortest distance from each first vocabulary vector of the social text data to all second vocabulary vectors of the first story line by using a cosine similarity measure method, and taking the average value of the shortest distances corresponding to all first vocabulary vectors as the similarity between the social text data and the first story line.
In this optional embodiment, the word vector model word2vec may be trained for vector operation in a vector space, and a computation logic of the word vector model word2vec is to determine similarity between social text data and each first story line by using a cosine similarity measure, which relates to a computation formula as follows:
wherein, word i First vocabulary of ith participle in social text dataVector, keyword j And the second vocabulary vector of the jth keyword in all the keyword libraries of the current first story line and the ith word segmentation is the shortest distance from the first vocabulary vector, and I is the word segmentation quantity of the social text data. The distatnce represents the graph distance, namely the similarity, between the current first story line and the social text data, and the numerical range is 0-1.
In this way, the similarity between each first story line and the social text data can be effectively determined, and the operation is rapid when the data volume is large based on the full matrix operation.
And S14, taking the first story line with the similarity meeting the preset threshold requirement as a target story line, and determining a target recommended product according to the target story line.
Specifically, the similarity of each first story line determined in step S13 is sorted from large to small, and then the first story line meeting the preset threshold requirement is determined as a target story line, so that a target recommended product can be determined. The preset threshold may be any value between 0.9 and 1, and may be specifically set according to an actual situation.
According to the technical scheme, a plurality of scene lines used for identifying user requirements are constructed, then the social text data of the user is semantically matched with each story node of each scene line, so that the first story lines which are matched relatively are selected, then word segmentation processing is carried out on the social text data, each word segmentation and the story label key words of each first story line are accurately filtered in a word vector measurement mode, the target story line is obtained, the target recommended product is determined, the product recommendation accuracy is effectively improved, and the marketing efficiency of follow-up marketing services is facilitated to be improved.
Referring to fig. 2, fig. 2 is a functional block diagram of a preferred embodiment of the social data based intelligent product recommendation apparatus according to the present invention. The intelligent product recommendation device 11 based on social data includes a construction scenario unit 110, a matching data unit 111, a determination vector unit 112, a similarity calculation unit 113, and a product recommendation unit 114.
The scene constructing unit 110 is configured to construct a plurality of scene lines for identifying user requirements, where the scene lines include a plurality of story nodes, each story node includes a story label and a preset semantic classification model, and each scene line corresponds to at least one recommended product.
The matching data unit 111 is configured to obtain social text data of a user, determine a matching degree between the social text data and each scene line according to the semantic classification models of the scene lines, and use the scene line meeting the matching requirement as a first story line.
A vector determining unit 112, configured to determine a first vocabulary vector of each word in the social text data, and determine a second vocabulary vector of each preset keyword of the story tag in each first story line.
A similarity calculation unit 113, configured to determine a similarity between the social text data and the first story line according to all first vocabulary vectors corresponding to the social text data and all second vocabulary vectors corresponding to each of the first story lines.
And the product recommending unit 114 is used for taking the first story line with the similarity meeting the preset threshold requirement as a target story line and determining a target recommended product according to the target story line.
According to the technical scheme, a plurality of scene lines for identifying user requirements are established, then the social text data of the user is semantically matched with each story node of each scene line, so that the first story lines which are matched relatively are selected, then word segmentation processing is carried out on the social text data, each word segmentation and the story label key words of each first story line are accurately filtered in a word vector measurement mode, the target story line is obtained, the target recommended product is determined, the product recommendation accuracy is effectively improved, and the marketing efficiency of follow-up marketing services is improved.
The specific limitations of the steps of the intelligent product recommendation device based on social data can be referred to the limitations of the intelligent product recommendation method based on social data, and are not described herein again. Furthermore, it should be noted that, all or part of the modules in the social data based intelligent product recommendation device may be implemented by software, hardware and a combination thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.
Fig. 3 is a schematic structural diagram of an electronic device according to an embodiment of the present disclosure. The electronic device 1 comprises a memory 12 and a processor 13. The memory 12 is used for storing computer readable instructions, and the processor 13 is used for executing the computer readable instructions stored in the memory to implement the intelligent product recommendation method based on social data according to any one of the above embodiments.
In an alternative embodiment, the electronic device 1 further comprises a bus, a computer program stored in said memory 12 and executable on said processor 13, such as a social data based intelligent product recommendation program.
Fig. 3 shows only the electronic device 1 with the memory 12 and the processor 13, and it will be understood by those skilled in the art that the structure shown in fig. 3 does not constitute a limitation of the electronic device 1, and may comprise fewer or more components than shown, or a combination of certain components, or a different arrangement of components.
In conjunction with fig. 1, the memory 12 in the electronic device 1 stores a plurality of computer-readable instructions to implement a method for intelligent product recommendation based on social data, and the processor 13 can execute the plurality of instructions to implement:
constructing a plurality of scene lines for identifying user requirements, wherein the scene lines comprise a plurality of story nodes, each story node comprises a story label and a preset semantic classification model, and each scene line corresponds to at least one recommended product;
obtaining social text data of a user, determining the matching degree of the social text data and each scene line according to the semantic classification models of the scene lines, and taking the scene lines meeting the matching requirements as first story lines;
determining a first vocabulary vector of each participle in the social text data, and determining a second vocabulary vector of each story tag preset keyword in each first story line;
determining the similarity between the social text data and the first story line according to all first vocabulary vectors corresponding to the social text data and all second vocabulary vectors corresponding to each first story line;
and taking the first story line with the similarity meeting the preset threshold requirement as a target story line, and determining a target recommended product according to the target story line.
Specifically, the processor 13 may refer to the description of the relevant steps in the embodiment corresponding to fig. 1 for a specific implementation method of the instruction, which is not described herein again.
It will be understood by those skilled in the art that the schematic diagram is only an example of the electronic device 1, and does not constitute a limitation to the electronic device 1, the electronic device 1 may have a bus-type structure or a star-shaped structure, the electronic device 1 may further include more or less hardware or software than those shown in the figures, or different component arrangements, for example, the electronic device 1 may further include an input and output device, a network access device, etc.
It should be noted that the electronic device 1 is only an example, and other existing or future electronic products, such as may be adapted to the present application, should also be included in the scope of protection of the present application, and is included by reference.
The processor 13 may be composed of an integrated circuit in some embodiments, for example, a single packaged integrated circuit, or may be composed of a plurality of integrated circuits packaged with the same function or different functions, including one or more Central Processing Units (CPUs), microprocessors, digital Processing chips, graphics processors, and combinations of various control chips. The processor 13 is a Control Unit (Control Unit) of the electronic device 1, connects various components of the electronic device 1 by using various interfaces and lines, and executes various functions and processes data of the electronic device 1 by running or executing programs or modules (for example, executing a social data-based intelligent product recommendation program, etc.) stored in the memory 12 and calling data stored in the memory 12.
The processor 13 executes the operating system of the electronic device 1 and various installed application programs. The processor 13 executes the application program to implement the steps in each of the above embodiments of the social data based intelligent product recommendation method, such as the steps shown in fig. 1.
Illustratively, the computer program may be partitioned into one or more modules/units, which are stored in the memory 12 and executed by the processor 13 to accomplish the present application. The one or more modules/units may be a series of computer readable instruction segments capable of performing certain functions, which are used to describe the execution of the computer program in the electronic device 1. For example, the computer program may be divided into a build scenario unit 110, a matching data unit 111, a determination vector unit 112, a similarity calculation unit 113, a product recommendation unit 114.
The integrated unit implemented in the form of a software functional module may be stored in a computer-readable storage medium. The software functional module is stored in a storage medium and includes several instructions to enable a computer device (which may be a personal computer, a computer device, or a network device) or a Processor (Processor) to execute parts of the social data based intelligent product recommendation method according to various embodiments of the present application.
The integrated modules/units of the electronic device 1 may be stored in a computer-readable storage medium if they are implemented in the form of software functional units and sold or used as separate products. Based on such understanding, all or part of the processes in the methods of the embodiments described above may be implemented by a computer program, which may be stored in a computer-readable storage medium and executed by a processor, to implement the steps of the embodiments of the methods described above.
Wherein the computer program comprises computer program code, which may be in the form of source code, object code, an executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U-disk, removable hard disk, magnetic disk, optical disk, computer Memory, read-Only Memory (ROM), random-access Memory and other Memory, etc.
Further, the computer-readable storage medium may mainly 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 can store scene lines, keywords of story labels, recommended products and other data.
The bus may be a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, or the like. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, only one arrow is shown in FIG. 3, but this does not indicate only one bus or one type of bus. The bus is arranged to enable connection communication between the memory 12 and at least one processor 13 etc.
The embodiment of the present application further provides a computer-readable storage medium (not shown), in which computer-readable instructions are stored, and the computer-readable instructions are executed by a processor in an electronic device to implement the social data based intelligent product recommendation method according to any of the above embodiments.
In the several embodiments provided in the present application, it should be understood that the disclosed system, apparatus and method may be implemented in other manners. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the modules is only one logical functional division, and other divisions may be realized in practice.
The modules described as separate parts may or may not be physically separate, and parts displayed as modules may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment.
In addition, functional modules in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, or in a form of hardware plus a software functional module.
Furthermore, it is obvious that the word "comprising" does not exclude other elements or steps, and the singular does not exclude the plural. A plurality of units or means stated in the description may also be implemented by one unit or means in software or hardware. The terms first, second, etc. are used to denote names, but not any particular order.
Finally, it should be noted that the above embodiments are only used for illustrating the technical solutions of the present application and not for limiting, and although the present application is described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications or equivalent substitutions can be made to the technical solutions of the present application without departing from the spirit and scope of the technical solutions of the present application.
Claims (10)
1. An intelligent product recommendation method based on social data, the method comprising:
constructing a plurality of scene lines for identifying user requirements, wherein the scene lines comprise a plurality of story nodes, each story node comprises a story label and a preset semantic classification model, and each scene line corresponds to at least one recommended product;
obtaining social text data of a user, determining the matching degree of the social text data and each scene line according to the semantic classification models of the scene lines, and taking the scene lines meeting the matching requirements as first story lines;
determining a first vocabulary vector of each participle in the social text data, and determining a second vocabulary vector of each story tag preset keyword in each first story line;
determining the similarity between the social text data and the first story line according to all first vocabulary vectors corresponding to the social text data and all second vocabulary vectors corresponding to each first story line;
and taking the first story line with the similarity meeting the preset threshold requirement as a target story line, and determining a target recommended product according to the target story line.
2. The social data based intelligent product recommendation method of claim 1 wherein the semantic classification model comprises a two-classification model; the determining the matching degree of the social text data and each scene line according to the semantic classification models of the scene lines comprises:
inputting the social text data into each two classification models of the scene line for classification to obtain matching probability output by each two classification model and using the matching probability as matching probability of a corresponding story node;
and performing joint probability distribution operation on the matching probabilities of all story nodes of the scene line to obtain the matching probability of the scene line and social text data, and taking the matching probability as the matching degree.
3. The social data based intelligent product recommendation method according to claim 2, wherein the two classification models comprise a Bert model and an integrated classifier connected with the rear end of the Bert model, and the matching probability of the story node is an average value of classification probabilities output by all sub-classifiers in the integrated classifier.
4. The social data based intelligent product recommendation method of claim 3, wherein the training process of the Bert model comprises:
obtaining a social text training data set, wherein samples of the social text training data set are marked with positive/negative sample categories;
inputting a social text training data set into a Bert model for coding, outputting a prediction classification result through a rear-end integrated classifier, comparing the prediction classification result with the positive/negative sample categories of corresponding samples, and determining a prediction error;
and adjusting model parameters of the Bert model according to the prediction error, performing gradient descent learning on a coding part of the Bert model, performing ensemble learning on a rear-end ensemble classifier, and performing iterative training until the prediction error is smaller than a preset error threshold value to obtain the completely trained Bert model.
5. An intelligent social data-based product recommendation method as claimed in claim 1, wherein each story tag is preset with at least one keyword; the determining a first vocabulary vector of each participle in the social text data and determining a second vocabulary vector of each preset keyword of the story tag in each first story line comprises the following steps:
performing Word segmentation on the social text data, and converting each Word segmentation into a corresponding first Word vector by using a preset Word vector model Word2 vec;
and converting all preset keywords of each story label in the first story line into corresponding second Word vectors by using a preset Word vector model Word2 vec.
6. The social data based smart product recommendation method of claim 1 wherein determining the similarity between the social text data and the first story line based on all first vocabulary vectors corresponding to the social text data and all second vocabulary vectors corresponding to each of the first story lines comprises:
and determining the shortest distance from each first vocabulary vector of the social text data to all second vocabulary vectors of the first story line by using a cosine similarity measure method, and taking the average value of the shortest distances corresponding to all first vocabulary vectors as the similarity between the social text data and the first story line.
7. The social data based intelligent product recommendation method of claim 1, wherein the recommended product comprises at least one service product in at least one financial topic.
8. An intelligent product recommendation device based on social data, the device comprising:
the system comprises a scene building unit, a recommendation unit and a recommendation unit, wherein the scene building unit is used for building a plurality of scene lines for identifying user requirements, each scene line comprises a plurality of story nodes, each story node comprises a story label and a preset semantic classification model, and each scene line corresponds to at least one recommended product;
the matching data unit is used for acquiring social text data of a user, determining the matching degree of the social text data and each scene line according to the semantic classification models of the scene lines, and taking the scene lines meeting the matching requirements as first story lines;
the vector determining unit is used for determining a first vocabulary vector of each participle in the social text data and determining a second vocabulary vector of each story label preset keyword in each first story line;
the similarity calculation unit is used for determining the similarity between the social text data and the first story line according to all first vocabulary vectors corresponding to the social text data and all second vocabulary vectors corresponding to each first story line;
and the product recommendation unit is used for taking the first story line with the similarity meeting the preset threshold requirement as a target story line and determining a target recommended product according to the target story line.
9. An electronic device, characterized in that the electronic device comprises:
a memory storing computer readable instructions; and
a processor executing computer readable instructions stored in the memory to implement the social data based intelligent product recommendation method of any one of claims 1 to 7.
10. A computer-readable storage medium having computer-readable instructions stored thereon, which when executed by a processor implement the social data based intelligent product recommendation method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210992127.0A CN115358817A (en) | 2022-08-17 | 2022-08-17 | Intelligent product recommendation method, device, equipment and medium based on social data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210992127.0A CN115358817A (en) | 2022-08-17 | 2022-08-17 | Intelligent product recommendation method, device, equipment and medium based on social data |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115358817A true CN115358817A (en) | 2022-11-18 |
Family
ID=84002638
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210992127.0A Pending CN115358817A (en) | 2022-08-17 | 2022-08-17 | Intelligent product recommendation method, device, equipment and medium based on social data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115358817A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117093783A (en) * | 2023-04-12 | 2023-11-21 | 浙江卡赢信息科技有限公司 | Intelligent recommendation system and method for point exchange combined with user social data |
-
2022
- 2022-08-17 CN CN202210992127.0A patent/CN115358817A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117093783A (en) * | 2023-04-12 | 2023-11-21 | 浙江卡赢信息科技有限公司 | Intelligent recommendation system and method for point exchange combined with user social data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108597519B (en) | Call bill classification method, device, server and storage medium | |
| CN111695033B (en) | Enterprise public opinion analysis method, enterprise public opinion analysis device, electronic equipment and medium | |
| CN107808011B (en) | Information classification extraction method and device, computer equipment and storage medium | |
| CN112613308B (en) | User intention recognition method, device, terminal equipment and storage medium | |
| WO2022105115A1 (en) | Question and answer pair matching method and apparatus, electronic device and storage medium | |
| CN109086265B (en) | Semantic training method and multi-semantic word disambiguation method in short text | |
| WO2021204017A1 (en) | Text intent recognition method and apparatus, and related device | |
| CN109299227B (en) | Information query method and device based on voice recognition | |
| CN112926308B (en) | Method, device, equipment, storage medium and program product for matching text | |
| CN112287069A (en) | Information retrieval method and device based on voice semantics and computer equipment | |
| CN111401065A (en) | Entity identification method, device, equipment and storage medium | |
| CN113326702A (en) | Semantic recognition method and device, electronic equipment and storage medium | |
| CN115358817A (en) | Intelligent product recommendation method, device, equipment and medium based on social data | |
| WO2021217866A1 (en) | Method and apparatus for ai interview recognition, computer device and storage medium | |
| CN111783425B (en) | Intention identification method based on syntactic analysis model and related device | |
| CN117520503A (en) | Financial customer service dialogue generation method, device, equipment and medium based on LLM model | |
| CN112256863A (en) | Method and device for determining corpus intentions and electronic equipment | |
| CN111783424A (en) | Text clause dividing method and device | |
| CN116402166A (en) | Training method and device of prediction model, electronic equipment and storage medium | |
| WO2023137903A1 (en) | Reply statement determination method and apparatus based on rough semantics, and electronic device | |
| CN116090450A (en) | Text processing method and computing device | |
| CN115510188A (en) | Text keyword association method, device, equipment and storage medium | |
| US20220382982A1 (en) | System and method of automatic topic detection in text | |
| CN114626463A (en) | Language model training method, text matching method and related device | |
| CN114398482A (en) | Dictionary construction method and device, electronic equipment and storage medium |
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 |