CN110738545A - Product recommendation method and device based on user intention identification, computer equipment and storage medium - Google Patents
Product recommendation method and device based on user intention identification, computer equipment and storage medium Download PDFInfo
- Publication number
- CN110738545A CN110738545A CN201910814684.1A CN201910814684A CN110738545A CN 110738545 A CN110738545 A CN 110738545A CN 201910814684 A CN201910814684 A CN 201910814684A CN 110738545 A CN110738545 A CN 110738545A
- Authority
- CN
- China
- Prior art keywords
- user
- information
- candidate
- candidate user
- intention
- 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
- 238000000034 method Methods 0.000 title claims abstract description 62
- 238000004891 communication Methods 0.000 claims description 30
- 230000008569 process Effects 0.000 claims description 25
- 239000013598 vector Substances 0.000 claims description 23
- 238000003066 decision tree Methods 0.000 claims description 22
- 238000013507 mapping Methods 0.000 claims description 13
- 230000004044 response Effects 0.000 claims description 7
- 238000007781 pre-processing Methods 0.000 claims description 5
- 230000000717 retained effect Effects 0.000 claims description 5
- 150000001875 compounds Chemical class 0.000 claims description 3
- 201000010099 disease Diseases 0.000 description 7
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 7
- 238000001514 detection method Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 230000036541 health Effects 0.000 description 4
- 238000004590 computer program Methods 0.000 description 3
- 238000013461 design Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000014509 gene expression Effects 0.000 description 3
- 238000012549 training Methods 0.000 description 3
- 230000006399 behavior Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 230000015572 biosynthetic process Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 238000005065 mining Methods 0.000 description 2
- 238000003786 synthesis reaction Methods 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000004931 aggregating effect Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000012790 confirmation Methods 0.000 description 1
- 230000010485 coping Effects 0.000 description 1
- 208000029078 coronary artery disease Diseases 0.000 description 1
- 230000003862 health status Effects 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
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
- 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
Landscapes
- Business, Economics & Management (AREA)
- Finance (AREA)
- Engineering & Computer Science (AREA)
- Accounting & Taxation (AREA)
- Marketing (AREA)
- Economics (AREA)
- Development Economics (AREA)
- Strategic Management (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Technology Law (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application relates to product recommendation methods and devices based on user intention identification, computer equipment and storage media.
Description
Technical Field
The invention relates to the technical field of intelligent robots, in particular to product recommendation methods, devices, computer equipment and storage media based on user intention identification.
Background
For example, external recommenders of insurance companies such as business salesmen of the bank insurance business lack system insurance knowledge, cannot professionally serve the customers, influence the sales success rate, and have high labor cost and management cost for serving the seats and customer services of the external recommenders.
Disclosure of Invention
The embodiment of the application provides product recommendation methods, devices, computer equipment and storage media based on user intention identification, and can solve at least part of the problems.
The embodiment of the application provides product recommendation methods based on user intention identification, which comprise the following steps:
acquiring user information of candidate users;
determining a recommended product corresponding to the candidate user according to the user information;
selecting a phone book corresponding to the recommended product from a pre-constructed phone book library, wherein the phone book comprises a plurality of communication link contents, the communication link contents aim at determining whether a user has purchase intention, the communication link contents are connected in series in a decision tree mode, the decision tree is formed by connecting nodes corresponding to all steps of the communication link contents, and other nodes except the last nodes are provided with access conditions for entering the next nodes;
and calling the candidate user, determining whether the candidate user has purchase intention or not according to the phone script during the call, and marking the candidate user as a target user if the candidate user is determined to have the purchase intention.
The embodiment of the application provides product recommendation devices based on user intention identification, which include:
the information acquisition module is used for acquiring the user information of the candidate user;
the product determining module is used for determining recommended products corresponding to the candidate users according to the user information;
the phone book determination module is used for selecting a phone book corresponding to the recommended product from a pre-constructed phone book library, wherein the phone book comprises a plurality of communication link contents, the plurality of communication link contents are used for determining whether a user has a purchase intention, the plurality of communication link contents are connected in series in a decision tree mode, the decision tree is formed by connecting nodes corresponding to all steps of the plurality of communication link contents, and other nodes except the last nodes are provided with access conditions for entering the next nodes;
and the intelligent outbound module is used for calling the candidate user, determining whether the candidate user has purchase intention or not according to the phone script in a call, and marking the candidate user as a target user if the candidate user is determined to have the purchase intention.
The embodiment of the present application further provides computer devices, which include 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 product recommendation method identified based on the user intention.
Embodiments of the present application also provide a storage medium storing computer readable instructions that, when executed by or more processors, cause or more processors to perform the steps of the above-described method for identifying product recommendations based on user intent.
According to the product recommendation method, the product recommendation device, the computer equipment and the storage medium based on the user intention identification, the corresponding recommended product is determined according to the user information, the telephone book corresponding to the recommended product is selected from the telephone book library, then the user is called, communication is carried out in the communication according to the selected telephone book, and whether the user has the purchase intention is determined. In the whole process, manual calling is not needed, so that the labor cost is saved, and the working efficiency is improved. Moreover, the dialect book has related introduction contents of recommended products, is a dialect book with professional knowledge, and can serve users professionally in the conversation process, so that the sale success rate is improved.
Drawings
FIG. 1 is a block diagram showing the internal structure of a computer device in embodiments;
FIG. 2 is a flow diagram of a method for identifying product recommendations based on user intent in embodiments;
FIG. 3 is a flow chart of identifying user intent in embodiments;
FIG. 4 is a block diagram of exemplary product recommendation devices identified based on user intent.
Detailed Description
For purposes of making the objects, aspects and advantages of the present invention more apparent, the present invention will be described in detail below with reference to the accompanying drawings and examples.
It will be understood that the terms "", "second", etc. as used herein may be used herein to describe various elements, but these elements are not limited by these terms.
FIG. 1 is a schematic diagram of a computer device according to embodiments of the present application, which may be an intelligent robot since the method for recommending a product based on user intent recognition shown in FIG. 2 may be performed by the intelligent robot, FIG. 1 shows that the computer device includes a processor, a non-volatile storage medium, a memory, and a network interface connected through a system bus, wherein the non-volatile storage medium of the computer device stores an operating system, a database, and computer readable instructions, which may store a sequence of control information, and which, when executed by the processor, may cause the processor to implement methods for recommending a product based on user intent recognition.
The embodiment of the present application provides product recommendation methods based on user intention identification, where the method is executed by the computer device provided in fig. 1, and the specific form of the computer device may be an intelligent robot, as shown in fig. 2, the product recommendation method based on user intention identification provided in this embodiment includes the following steps:
s210, acquiring user information of a candidate user;
it will be appreciated that the products may be insurance products, loan products, and of course other products. The user information that needs to be known is different when recommending different products. The following is presented in two cases:
(1) when the recommended product is a loan product, the user information to be known may include: the user information comprises historical transaction data of the candidate user and a financial institution, historical operation data of the candidate user on a preset application program, personal positioning data of the candidate user, personal attribute data of the candidate user and the like. The historical transaction data includes, for example, whether the candidate user has logged out halfway in applying for the loan, whether the candidate user has credit card billing, and the like. And the candidate user presets historical operation data of the application program, for example, whether the candidate user frequently logs in a certain loan software, the login time, the use duration and the like. The personal positioning data of the candidate user, for example, the positioning data of the candidate user at each time point, the position where the candidate user appears during working hours is regarded as the working address of the candidate user, and the position where the candidate user appears during sleeping hours at night is regarded as the living address of the candidate user. Personal attribute data of the candidate user, for example, age group, gender, marital status, health status, child care status, income status, and the like. The user information needs to be acquired by the intelligent robot from a financial institution, a mobile terminal of the user, a server of a preset application program and the like through a network.
(2) When the recommended product is an insurance product, the user information to be known may include: age group, gender, occupation, income, health (with or without significant disease history), etc. The user information at this time corresponds to the personal attribute data in the above.
Of course, for other products, when other user information needs to be known, in case, when the intelligent robot makes a recommendation, the user information needs to be known is at least items of historical transaction data with the financial institution, historical operation data of a preset application program, personal positioning data and personal attribute data.
In a practical scenario, the intelligent robot may perform remote recommendation or field recommendation when performing recommendation, and for remote recommendation, the intelligent robot may communicate with a remote user by calling out, as shown in steps S230 and S240, the remote recommendation may be performed by calling out, for a field user, product recommendation may be performed by displaying an interface or introducing a field voice, and in this case, the specific process of obtaining user information of a candidate user may include providing an information filling interface in response to an preset operation on the intelligent robot, obtaining user information filled on the information filling interface in response to an information submitting operation after the information filling operation on the information filling interface, the candidate user may be a field user, that is, the user performs a certain operation on the intelligent robot, triggering the intelligent robot to display the information filling interface, performing information submission after the user information filling on the information filling interface, and the intelligent robot may obtain user information on the information filling interface in response to the submitting operation , the user information filling method may be applied to obtain information of health information required by users (such as information required by a simple operation, information providing a history of health information, and information required by a user, such as a history of a user, and a history of providing a history of a user, and a history of a user, which may not be accurate insurance, and a history of providing a history of a user, and a history of a user.
S220, determining a recommended product corresponding to the candidate user according to the user information;
in practical applications, the corresponding recommended product may be determined according to the user information in various ways, wherein ways are:
s221a, determining the feature label of the candidate user according to the user information;
s222a, determining the client group to which the candidate user belongs according to the preset matching rule and the feature tag of each client group;
for example, the preset matching rules of client groups, namely "Shenzhen handroom mortuary loan intention client group" include three conditions, namely, a region is Shenzhen, b no real estate, and c, the house purchasing consultation app is frequently used in the last 1 year, and the three conditions are ' AND ' relationship, only the users meeting the three conditions are classified into the "Shenzhen handroom mortuary loan intention client group ', and at this time, whether the tags corresponding to the three conditions are included in the feature tags of the users needs to be checked, and whether the users are allocated to the client groups is determined.
It is understood that the preset matching rules of each client group need to be configured on the configuration platform of the intelligent robot before the step is executed, specifically, variables are selected from a variable library, and then corresponding condition rules are edited or configured as required, wherein the variables are regions, gender, age, presence or absence of real estate, frequent use of house-purchasing consultation apps, over-loan behavior, over-credit card billing staging behavior, and the like, and the condition rules are 'and', 'or', '' thereof, and the like.
S223a, selecting a recommended product corresponding to the customer group to which the candidate user belongs according to a pre-established mapping relation table; wherein the mapping relation table comprises mapping relations between the plurality of customer groups and a plurality of recommended products;
it can be understood that the mapping relationship table stores the mapping relationship between the customer group and the recommended product, and the mapping relationship table is configured on the configuration platform of the intelligent robot before the step is performed, specifically, after the preset matching rule of the customer group is created, a product suitable for the customer group is selected from the product library, and the product and the customer group form the mapping relationship, so that the product can be recommended to the user belonging to the customer group.
Another ways of determining the corresponding recommended product according to the user information are:
s221b, determining the feature label of the candidate user according to the user information;
s222b, selecting recommended products corresponding to the candidate users from the products according to respective preset matching rules and the feature labels of the products;
here, the corresponding matching rule is directly set for the product, the preset matching rule of the product is directly matched with the feature tag of the user, and then the product suitable for the user is determined. Therefore, before executing the step, the matching conditions of each product need to be configured on the configuration platform of the intelligent robot. For example, for a common insurance product, it is only necessary that the age group is matched, and the matching rule of the common insurance product is the age group. For insurance products with more strict requirements, the annual income, the age group and the health condition may be required to be met, for example, the annual income is over 20 ten thousand, the age group is 18-60 years, and no major disease history exists, so the insurance products are purchased, the matching rule is that the annual income is over 20 ten thousand, the age group is 18-60 years, and no major disease history exists, and the relationship among the three conditions is 'and'.
In practical applications, any ways may be selected to determine the recommended product, but other ways may be used to determine the recommended product.
S230, selecting a word-operation book corresponding to the recommended product from a pre-constructed word-operation library;
the phone book comprises a plurality of communication link contents, the communication link contents aim at determining whether a user has purchase intention, the communication link contents are connected in series in a decision tree mode, the decision tree is formed by connecting nodes corresponding to all steps of the communication link contents, and admission conditions for entering the next nodes are arranged on other nodes except the last nodes.
In practical applications, the communication link contents can include links of user intention detection, product introduction, question answering, manual visit reservation and the like, for example, for loan products, a conversational design can be performed in advance, answers to questions that can be asked under various possible situations, such as annual interest rate of loan, too low amount of loan and the like, are designed, coping strategies under various abnormal situations, such as no good signal, no question to the client and the like, are designed, then, the conversational of each step of each link is concatenated by using a decision tree, so that the robot answers under various situations enter different conversational branch flows according to the user answers, the designed conversational design text includes client intention detection (including saving clients who do not interest in the replies), product introduction, question answering, success answering, and finish words, the possible steps include a, client intention detection-client confirmation of interest-product-answer-reservation-success-finish-end words, b, client intention detection-client leave-answer words, intelligent call answering by a, a decision tree is performed, the machine answers are performed according to the qualitative question answering process of the question, the machine answers, the question answering process is performed, the question answering is performed according to the intelligent answer, the process of the question answering, the question answering process of the question answering is performed, the process of the intelligent answer is performed, and the process of.
It can be understood that a decision tree is an expression form of a conversational script, the decision tree is a -node network (formed by connecting a plurality of nodes), admission conditions are set on the nodes, and can be used as conditions for entering a lower node, the decision tree can also be called a conversational decision library or a conversational tree, and the like.
Of course, recommended products can be provided with different dialogs corresponding to users with different age groups and/or professional types, for example, if the users are white-collar users, the dialogs used in calling out are simple, and if the users are retired people, the dialogs with more figures, examples and the like are used in calling out to help the clients to understand.
S240, calling the candidate user, determining whether the candidate user has purchasing intention or not according to the phone book in a call, and if the candidate user is determined to have purchasing intention, marking the candidate user as a target user.
The intelligent outbound platform comprises three parts of voice recognition, conversation technology and voice synthesis, wherein the intelligent outbound platform converts voice replied by a user into characters through the voice recognition technology for subsequent processing and analysis, b, recognizes the content replied by the client through the conversation technology, namely, endows the intelligent robot with the capability of understanding the user intention, and can match pre-trained dialogs from a dialogs book, wherein is required to realize a dialogs decision tree obtained based on the training, and c, the matched dialogs are converted into voice through the voice synthesis mode, so that the intelligent robot realizes the voice communication with the user.
The method is mainly applied to loan recommendation, and comprises five link contents including intention detection, product introduction, question reply, reservation success and end words, wherein each link content can be divided into a plurality of steps, such as the steps of 'intention detection' including two steps of 'greeting for a client' and 'asking if the client is interested in', the two steps have a sequence, the steps correspond to nodes in a decision tree, and the nodes are connected with the next nodes.
In practical applications, the determining whether the candidate user has the purchase intention during the call according to the call script may include determining whether the candidate user has the purchase intention through multiple rounds of conversations according to the call script, wherein each round of conversation process includes:
s241, receiving th voice information replied by the candidate user;
s242, converting the th voice message into a corresponding th character message;
s243, identifying the user intention from the th character information;
s244, determining lower nodes to enter according to the user intention and the admission conditions of the corresponding nodes in the decision tree, and determining second character information for replying the candidate user according to the lower nodes;
and S245, converting the second text message into a corresponding second voice message for answering.
It is understood that after determining whether the user has the purchasing intention, a person can be arranged to visit the user with the purchasing intention, and of course, the purchasing link of the recommended product can be obtained and then sent to the mobile terminal of the target user, so that the labor cost can be reduced. The method firstly determines whether the user has the purchasing intention, and then sends the purchasing link to the user with the purchasing intention is targeted information delivery compared with the method of directly sending the purchasing intention to all users, so that the situation that a plurality of users without the purchasing intention are marked as promotion telephones or recommended telephones and further the users with the purchasing intention are influenced can be avoided.
For example, after the field user fills user information in the interface of the intelligent robot, and then the intelligent robot determines the recommended product corresponding to the candidate user according to the user information, a product recommendation interface is provided , and the product recommendation interface displays introduction content and a purchase link of the recommended product.
Of course, if the user still has questions to be consulted through the introduction content on the interface, the intelligent robot can be consulted for relevant content in a voice manner, and the specific consultation process includes:
receiving third voice information of the candidate user, wherein the third voice information is voice information of a problem of the field user on the recommended product; converting the third voice information into corresponding third text information; identifying user intention from the third text information; according to the user intention, performing entity matching in a knowledge graph corresponding to the recommended product to obtain a matching entity; determining fourth text information for replying the field user according to the matching entity and other entities connected with the matching entity; and converting the fourth text information into fourth voice information for answering. That is to say, the intelligent robot converts the voice of the user for question consultation into characters, recognizes the user intention according to the characters, performs entity matching in the knowledge graph according to the user intention, determines reply characters according to the matched entities, converts the reply characters into voice for playing, and accordingly answers the question of the user for consultation.
A knowledge graph as referred to herein is a product-specific knowledge graph, e.g., an insurance product-specific knowledge graph, and products are knowledge graphs.
The process of building the product's knowledge graph includes step 1 of schema design, i.e., defining which concepts, which attributes (e.g., female) or relationships (e.g., inclusion), such as those of insurance, duration, suitable population, etc., step 2 of vocabulary mining, including mining various synonyms, shorthand words, phrases, etc., and aggregating the synonyms into concepts, wherein the concepts refer to "entities" in the knowledge graph, i.e., entity discovery, specifically including entity implementation, entity classification, entity linking (i.e., words of different expressions are expressed using concepts and are represented by entities, e.g., premium, insurance cost, insurance amount, etc.), step 3 of relationship extraction, i.e., entity and entity are linked by relationships, such as "disease-specific" entity in the risk and "coronary" disease inclusion ", i.e., a specific disease in the risk contains this coronary disease 54 disease, step 4 of relationship extraction, since the knowledge graph data from different sources may be aligned with each other data, and the data from the same source is expressed using no more than 3675 years of the same data, or similar to the same age, and the same data (e., the same age, such as 3675).
It can be seen that the knowledge graph does not have a specific task, but the knowledge of products is embodied in the form of a network composed of points and edges, so the knowledge graph is only expression forms of knowledge, when a problem consultation is carried out, the intelligent robot matches corresponding entities in the knowledge graph and converts the replied content into voice output based on the understanding of the intention of a user, namely, identifies problems in the communication content, identifies entities involved in the identified problems, matches the entities in the knowledge graph, and determines the replied content according to the matched entities and other entities connected with the entities.
For example, when the user asks for the problem of premium calculation, according to the relevant data of the insured who is input by the user in the conversation process, the intelligent robot matches the corresponding entity-the premium estimation model suitable for the insured in the knowledge map according to the relevant data input by the user, and the premium is calculated by the estimation model and then output to the user. It is understood that there are multiple entities corresponding to the calculation model in the knowledge-graph.
Of course, different knowledge maps can be set for recommended products, corresponding to users of different age groups and/or professional types, for example, if the users are white-collar, the languages in the knowledge maps are more concise, and if the users are retired people, the knowledge maps are used with more figures, examples and the like, which are helpful for the understanding of the clients.
As shown in FIG. 3, the process of identifying the user intention can comprise the steps of preprocessing the text information, converting the preprocessed text information into a plurality of word vectors, updating the word vectors through a long-short term memory network, extracting text features from the updated word vectors, and obtaining corresponding intention categories according to the text features, wherein the word vectors are updated through a formula in the long-short term memory network, and the formula comprises the following steps:
in the formula (I), the compound is shown in the specification,
to update the t-th word vector before, Ct-1For the t-1 th word vector after update, ftIs Ct-1Of the retained proportionality coefficient itIs composed of
The scale factor retained in (1).
The whole intention recognition process can be realized by an intention recognition model, and the training process of the intention recognition model can roughly comprise the steps of obtaining linguistic data required by training, preprocessing the linguistic data, generating word vectors according to the preprocessed linguistic data, extracting features in the word vectors by using LSTM, and performing intention category recognition by using Softmax based on the extracted features, wherein the first three steps are preprocessing stages, word vectors can be generated by using word2vec, namely, each word is mapped to each vector, and the vectors are used for expressing the relation between words.
The main idea of LSTM is to pass each word in the sentence from the current to the next steps, i.e. to construct a language model to predict the meaning, part of speech, etc. of the next words based on previous words, LSTM adds a "processor" in the algorithm to determine if the information is useful before extracting features, to analyze if the information is useful, to discard useless information and useful information after updating, it can be seen that the process flow of LSTM is to input, forget and update information, extract information features, so that Softmax makes the final multi-classification.
Wherein, the role of Softmax is mainly to map the features of the input to real numbers in the range of [0,1], and ensure the sum of all classes to be 1, i.e. the probability of all classes is exactly equal to 1. In the process of intention identification, different intentions need to be classified, and the most possible intention of the user is judged according to the probability. The formula for calculating the probability for each class can be expressed as:
in the formula, i represents the ith element in the classification array V, and the softmax value of the element is the ratio of the index of the element to the sum of indexes of all elements.
According to the intelligent recommendation method provided by the embodiment of the application, the corresponding recommended products are determined according to the user information, then the dialect book corresponding to the recommended products is selected in the dialect library, then the user is called, communication is carried out according to the selected dialect book in the call, and then whether the user has the purchase intention is determined. In the whole process, manual calling is not needed, so that the labor cost is saved, and the working efficiency is improved. Moreover, the dialect book has related introduction contents of recommended products, is a dialect book with professional knowledge, and can serve users professionally in the conversation process, so that the sale success rate is improved.
As shown in FIG. 4, in embodiments, user-intent-based product recommendation devices 400 are provided, and the user-intent-based product recommendation device 400 may be integrated with the computer apparatus shown in FIG. 1, and the device 400 may specifically include:
an information obtaining module 410, configured to obtain user information of a candidate user;
a product determining module 420, configured to determine, according to the user information, a recommended product corresponding to the candidate user;
a phone book determining module 430, configured to select a phone book corresponding to the recommended product from a pre-constructed phone book library, where the phone book includes a plurality of communication link contents, the plurality of communication link contents are for determining whether a user has a purchase intention, the plurality of communication link contents are connected in series in a decision tree, the decision tree is formed by connecting nodes corresponding to respective steps of the plurality of communication link contents, and other nodes except for the last nodes are provided with admission conditions for entering the next nodes;
the intelligent outbound module 440 is configured to call the candidate user, determine whether the candidate user has an intention to purchase according to the phone book during a call, and mark the candidate user as a target user if it is determined that the candidate user has the intention to purchase.
In embodiments, the apparatus further comprises:
and the link pushing module is used for acquiring the purchase link of the recommended product and sending the purchase link to the mobile terminal of the target user.
In , the product determination module is specifically configured to determine a feature tag of the candidate user according to the user information, determine a customer group to which the candidate user belongs according to a preset matching rule and the feature tag of each of a plurality of customer groups, select a recommended product corresponding to the customer group to which the candidate user belongs according to a pre-established mapping relation table, where the mapping relation table includes mapping relations between the plurality of customer groups and the plurality of recommended products, or the product determination module is specifically configured to determine the feature tag of the candidate user according to the user information, and select the recommended product corresponding to the candidate user from the plurality of products according to the preset matching rule and the feature tag of each of the plurality of products, where the user information includes at least items of historical transaction data of the candidate user and a financial institution, historical operation data of the candidate user in a preset application program, personal positioning data of the candidate user, and personal attribute data of the candidate user.
In , the intelligent outbound module is specifically configured to determine whether the candidate user has an intention to purchase through multiple rounds of conversations according to the conversational script, wherein each round of conversation process includes receiving th speech information replied by the candidate user, converting the th speech information into corresponding th text information, identifying a user intention from the th text information, determining the next nodes to be entered according to the user intention and the admission conditions of the corresponding nodes in the decision tree, determining second text information replied to the candidate user according to the next nodes, and converting the second text information into corresponding second speech information to reply.
In , the information obtaining module is specifically configured to provide a information filling interface in response to a th preset operation on the intelligent robot, obtain user information filled on the information filling interface in response to an information submitting operation after the information filling operation on the information filling interface, wherein the candidate user is a live user, and the device further includes an information presentation module configured to provide a product recommendation interface on which introduction content and purchase links of the recommended product are presented.
In embodiments, the apparatus further comprises:
the product consultation module is used for receiving third voice information of the candidate user, wherein the third voice information is the voice information of the problem of the field user to the recommended product; converting the third voice information into corresponding third text information; identifying user intention from the third text information; according to the user intention, performing entity matching in a knowledge graph corresponding to the recommended product to obtain a matching entity; determining fourth text information for replying the field user according to the matching entity and other entities connected with the matching entity; and converting the fourth text information into fourth voice information for answering.
In , the process of recognizing the user's intention from the text information by the intelligent call-out module or the product consultation module includes preprocessing the text information, converting the preprocessed text information into a plurality of word vectors, updating the word vectors through the long-short term memory network, extracting text features from the updated word vectors, and obtaining corresponding intention categories according to the text features, wherein the word vectors are updated through a formula in the long-short term memory network, and the formula includes:
in the formula (I), the compound is shown in the specification,
to update the t-th word vector before, Ct-1For the t-1 th word vector after update, ftIs Ct-1Of the retained proportionality coefficient itIs composed of
The scale factor retained in (1).
It can be understood that, for the explanation, example, and beneficial effects of the product recommendation device identified based on the user intention provided in the present application, reference may be made to corresponding parts in the product recommendation method identified based on the user intention, and details are not described herein.
In embodiments, computer devices are provided, the computer devices including a memory, a processor, and a computer program stored on the memory and executable on the processor, the processor implementing the above-described product recommendation method based on user intent identification when executing the computer program.
It can be understood that, for the computer device provided in the present application, portions related to explanation, examples, and beneficial effects of the content may refer to corresponding portions in the product recommendation method identified based on the user intention, and details are not described here.
In embodiments, storage media storing computer readable instructions that, when executed by or more processors, cause or more processors to implement the above-described method of identifying product recommendations based on user intent are presented.
It is understood that the storage medium provided in the present application, portions of the explanation, examples, and benefits of the content thereof may refer to corresponding portions of the product recommendation method identified based on the user's intention, and details thereof are not repeated herein.
It will be understood by those skilled in the art that all or part of the processes in the methods of the above embodiments may be implemented by instructing relevant hardware through a computer program, which may be stored in a computer readable storage medium, and when the program is executed, the processes of the above embodiments of the methods may be included.
The technical features of the embodiments described above may be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the embodiments described above are not described, but should be considered as being within the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
The above-mentioned embodiments only express several embodiments of the present invention, and the description thereof is more specific and detailed, but not construed as limiting the scope of the present invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the inventive concept, which falls within the scope of the present invention. Therefore, the protection scope of the present patent shall be subject to the appended claims.
Claims (10)
1, A product recommendation method identified based on a user's intent, comprising:
acquiring user information of candidate users;
determining a recommended product corresponding to the candidate user according to the user information;
selecting a phone book corresponding to the recommended product from a pre-constructed phone book library, wherein the phone book comprises a plurality of communication link contents, the communication link contents aim at determining whether a user has purchase intention, the communication link contents are connected in series in a decision tree mode, the decision tree is formed by connecting nodes corresponding to all steps of the communication link contents, and other nodes except the last nodes are provided with access conditions for entering the next nodes;
and calling the candidate user, determining whether the candidate user has purchase intention or not according to the phone script during the call, and marking the candidate user as a target user if the candidate user is determined to have the purchase intention.
2. The method of claim 1, further comprising:
and acquiring a purchase link of the recommended product, and sending the purchase link to the mobile terminal of the target user.
3. The method of claim 1, wherein the determining the recommended products corresponding to the candidate users according to the user information comprises:
determining the feature tag of the candidate user according to the user information;
determining the client group to which the candidate user belongs according to the preset matching rule and the feature tag of each client group;
selecting a recommended product corresponding to a customer group to which the candidate user belongs according to a pre-established mapping relation table; wherein the mapping relation table comprises mapping relations between the plurality of customer groups and a plurality of recommended products;
or, the determining the recommended product corresponding to the candidate user according to the user information includes:
determining the feature tag of the candidate user according to the user information;
selecting a recommended product corresponding to the candidate user from the plurality of products according to respective preset matching rules of the plurality of products and the feature labels;
wherein the user information includes at least items of historical transaction data of the candidate user with a financial institution, historical operating data of the candidate user on a preset application, personal location data of the candidate user, and personal attribute data of the candidate user.
4. The method of claim 1, wherein determining whether the candidate user has an intent to purchase during the call based on the phone book comprises determining whether the candidate user has an intent to purchase through multiple rounds of conversations based on the phone book, wherein each round of conversation process comprises:
receiving th voice information replied by the candidate user, converting the th voice information into corresponding th text information, identifying user intention from the th text information, determining the lower nodes to enter according to the user intention and the admission condition of the corresponding node in the decision tree, determining the second text information replied to the candidate user according to the lower nodes, and converting the second text information into corresponding second voice information for replying.
5. The method as claimed in claim 1, wherein the obtaining user information of candidate users comprises providing an information filling interface in response to th preset operation on the intelligent robot, obtaining user information filled on the information filling interface in response to an information submitting operation after the information filling operation on the information filling interface, wherein the candidate users are field users;
correspondingly, after the recommended product corresponding to the candidate user is determined according to the user information, the method further comprises providing a product recommendation interface, wherein the introduction content and the purchase link of the recommended product are displayed on the product recommendation interface.
6. The method of claim 5, further comprising:
receiving third voice information of the candidate user, wherein the third voice information is voice information of a problem of the field user on the recommended product; converting the third voice information into corresponding third text information; identifying user intention from the third text information; according to the user intention, performing entity matching in a knowledge graph corresponding to the recommended product to obtain a matching entity; determining fourth text information for replying the field user according to the matching entity and other entities connected with the matching entity; and converting the fourth text information into fourth voice information for answering.
7. The method of claim 4 or 6, wherein the step of identifying the user intention from the text information comprises preprocessing the text information, converting the preprocessed text information into a plurality of word vectors, updating the word vectors through a long-short term memory network, extracting text characteristics from the updated word vectors, and obtaining corresponding intention categories according to the text characteristics, wherein the word vectors are updated through an formula in the long-short term memory network, and the formula comprises:
in the formula (I), the compound is shown in the specification,
to update the t-th word vector before, Ct-1For the t-1 th word vector after update, ftIs Ct-1Of the retained proportionality coefficient itIs composed of
Ratio of middle retentionThe coefficients are illustrated.
8, A product recommendation device identified based on a user intent, comprising:
the information acquisition module is used for acquiring the user information of the candidate user;
the product determining module is used for determining recommended products corresponding to the candidate users according to the user information;
the phone book determination module is used for selecting a phone book corresponding to the recommended product from a pre-constructed phone book library, wherein the phone book comprises a plurality of communication link contents, the plurality of communication link contents are used for determining whether a user has a purchase intention, the plurality of communication link contents are connected in series in a decision tree mode, the decision tree is formed by connecting nodes corresponding to all steps of the plurality of communication link contents, and other nodes except the last nodes are provided with access conditions for entering the next nodes;
and the intelligent outbound module is used for calling the candidate user, determining whether the candidate user has purchase intention or not according to the phone script in a call, and marking the candidate user as a target user if the candidate user is determined to have the purchase intention.
Computer apparatus, 9, , comprising a memory and a processor, the memory having stored therein computer readable instructions which, when executed by the processor, cause the processor to perform the steps of the method for identifying product recommendations based on user intent as claimed in any of claims 1 to 7, .
10, storage media storing computer readable instructions which, when executed by the one or more processors, cause the one or more processors to perform the steps of the method for identifying product recommendations based on user intent as claimed in any of claims 1-7, .
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910814684.1A CN110738545A (en) | 2019-08-30 | 2019-08-30 | Product recommendation method and device based on user intention identification, computer equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910814684.1A CN110738545A (en) | 2019-08-30 | 2019-08-30 | Product recommendation method and device based on user intention identification, computer equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110738545A true CN110738545A (en) | 2020-01-31 |
Family
ID=69267577
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910814684.1A Pending CN110738545A (en) | 2019-08-30 | 2019-08-30 | Product recommendation method and device based on user intention identification, computer equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110738545A (en) |
Cited By (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111401951A (en) * | 2020-03-13 | 2020-07-10 | 重庆锐云科技有限公司 | Virtual client construction method for user information transparentization under secret storage |
| CN111401998A (en) * | 2020-03-12 | 2020-07-10 | 重庆智慧之源科技有限公司 | Method, device and server for accurately recommending technology transfer intention customers |
| CN111464707A (en) * | 2020-03-30 | 2020-07-28 | 中国建设银行股份有限公司 | Outbound call processing method, device and system |
| CN111540466A (en) * | 2020-04-19 | 2020-08-14 | 周玉娟 | Big data based intelligent medical information pushing method and big data medical cloud platform |
| CN111563151A (en) * | 2020-05-07 | 2020-08-21 | 腾讯科技（深圳）有限公司 | Information acquisition method, session configuration device and storage medium |
| CN111724908A (en) * | 2020-06-28 | 2020-09-29 | 北京来也网络科技有限公司 | Epidemic situation investigation method and device based on robot process automation RPA |
| CN111897931A (en) * | 2020-06-24 | 2020-11-06 | 深圳追一科技有限公司 | Session setting method and apparatus, server, computer readable storage medium |
| CN111914077A (en) * | 2020-08-06 | 2020-11-10 | 平安科技（深圳）有限公司 | Customized speech recommendation method, device, computer equipment and storage medium |
| CN111914075A (en) * | 2020-08-06 | 2020-11-10 | 平安科技（深圳）有限公司 | Customer label determination method, device, equipment and medium based on deep learning |
| CN112035624A (en) * | 2020-09-11 | 2020-12-04 | 上海明略人工智能(集团)有限公司 | Text recommendation method and device and storage medium |
| CN112202977A (en) * | 2020-09-03 | 2021-01-08 | 深圳市彬讯科技有限公司 | Outbound processing method, device, equipment and medium for enterprise operation |
| CN112330411A (en) * | 2020-11-17 | 2021-02-05 | 中国平安财产保险股份有限公司 | Group product recommendation method and device, computer equipment and storage medium |
| CN112884083A (en) * | 2021-03-31 | 2021-06-01 | 中国工商银行股份有限公司 | Intelligent outbound call processing method and device |
| CN112967141A (en) * | 2021-03-30 | 2021-06-15 | 建信金融科技有限责任公司 | On-line construction method, device, medium and electronic equipment for asset transaction |
| CN113342964A (en) * | 2021-06-03 | 2021-09-03 | 云南大学 | Recommendation type determination method and system based on mobile service |
| CN113468420A (en) * | 2021-06-29 | 2021-10-01 | 杭州摸象大数据科技有限公司 | Method and system for recommending products |
| CN113763096A (en) * | 2020-12-14 | 2021-12-07 | 北京京东尚科信息技术有限公司 | Product shopping guide processing method, device, electronic equipment, medium and program product |
| WO2022000141A1 (en) * | 2020-06-28 | 2022-01-06 | 北京来也网络科技有限公司 | Epidemic investigation method employing robotic process automation (rpa), and device |
| CN114049224A (en) * | 2021-10-13 | 2022-02-15 | 北京博瑞彤芸科技股份有限公司 | Training method and system for insurance businessman |
| CN114971795A (en) * | 2022-05-27 | 2022-08-30 | 中国平安财产保险股份有限公司 | Insurance product recommendation method and device, computer equipment and storage medium |
| WO2022267176A1 (en) * | 2021-06-23 | 2022-12-29 | 未鲲(上海)科技服务有限公司 | Product recommendation method and apparatus based on artificial intelligence, and device and storage medium |
| CN117252995A (en) * | 2023-11-17 | 2023-12-19 | 深圳市加推科技有限公司 | Intelligent business card using method, intelligent sales system and storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108875034A (en) * | 2018-06-25 | 2018-11-23 | 湖南丹尼尔智能科技有限公司 | A kind of Chinese Text Categorization based on stratification shot and long term memory network |
| CN109285030A (en) * | 2018-08-29 | 2019-01-29 | 深圳壹账通智能科技有限公司 | Products Show method, apparatus, terminal and computer readable storage medium |
| CN109462707A (en) * | 2018-11-13 | 2019-03-12 | 平安科技（深圳）有限公司 | Method of speech processing, device and computer equipment based on automatic outer call system |
| CN109829039A (en) * | 2018-12-13 | 2019-05-31 | 平安科技（深圳）有限公司 | Intelligent chat method, device, computer equipment and storage medium |
| CN109961786A (en) * | 2019-01-31 | 2019-07-02 | 平安科技（深圳）有限公司 | Products Show method, apparatus, equipment and storage medium based on speech analysis |
| CN110033851A (en) * | 2019-04-02 | 2019-07-19 | 腾讯科技（深圳）有限公司 | Information recommendation method, device, storage medium and server |
-
2019
- 2019-08-30 CN CN201910814684.1A patent/CN110738545A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108875034A (en) * | 2018-06-25 | 2018-11-23 | 湖南丹尼尔智能科技有限公司 | A kind of Chinese Text Categorization based on stratification shot and long term memory network |
| CN109285030A (en) * | 2018-08-29 | 2019-01-29 | 深圳壹账通智能科技有限公司 | Products Show method, apparatus, terminal and computer readable storage medium |
| CN109462707A (en) * | 2018-11-13 | 2019-03-12 | 平安科技（深圳）有限公司 | Method of speech processing, device and computer equipment based on automatic outer call system |
| CN109829039A (en) * | 2018-12-13 | 2019-05-31 | 平安科技（深圳）有限公司 | Intelligent chat method, device, computer equipment and storage medium |
| CN109961786A (en) * | 2019-01-31 | 2019-07-02 | 平安科技（深圳）有限公司 | Products Show method, apparatus, equipment and storage medium based on speech analysis |
| CN110033851A (en) * | 2019-04-02 | 2019-07-19 | 腾讯科技（深圳）有限公司 | Information recommendation method, device, storage medium and server |
Cited By (29)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111401998A (en) * | 2020-03-12 | 2020-07-10 | 重庆智慧之源科技有限公司 | Method, device and server for accurately recommending technology transfer intention customers |
| CN111401998B (en) * | 2020-03-12 | 2023-06-30 | 重庆智慧之源科技有限公司 | Accurate recommendation method, device and server for technological transfer intention clients |
| CN111401951A (en) * | 2020-03-13 | 2020-07-10 | 重庆锐云科技有限公司 | Virtual client construction method for user information transparentization under secret storage |
| CN111464707A (en) * | 2020-03-30 | 2020-07-28 | 中国建设银行股份有限公司 | Outbound call processing method, device and system |
| CN111540466B (en) * | 2020-04-19 | 2021-01-15 | 深圳坐标软件集团有限公司 | Big data based intelligent medical information pushing method and big data medical cloud platform |
| CN111540466A (en) * | 2020-04-19 | 2020-08-14 | 周玉娟 | Big data based intelligent medical information pushing method and big data medical cloud platform |
| CN111563151A (en) * | 2020-05-07 | 2020-08-21 | 腾讯科技（深圳）有限公司 | Information acquisition method, session configuration device and storage medium |
| CN111563151B (en) * | 2020-05-07 | 2024-02-02 | 腾讯科技（深圳）有限公司 | Information acquisition method, session configuration method, device and storage medium |
| CN111897931A (en) * | 2020-06-24 | 2020-11-06 | 深圳追一科技有限公司 | Session setting method and apparatus, server, computer readable storage medium |
| CN111724908A (en) * | 2020-06-28 | 2020-09-29 | 北京来也网络科技有限公司 | Epidemic situation investigation method and device based on robot process automation RPA |
| WO2022000141A1 (en) * | 2020-06-28 | 2022-01-06 | 北京来也网络科技有限公司 | Epidemic investigation method employing robotic process automation (rpa), and device |
| CN111914075A (en) * | 2020-08-06 | 2020-11-10 | 平安科技（深圳）有限公司 | Customer label determination method, device, equipment and medium based on deep learning |
| CN111914077A (en) * | 2020-08-06 | 2020-11-10 | 平安科技（深圳）有限公司 | Customized speech recommendation method, device, computer equipment and storage medium |
| CN112202977A (en) * | 2020-09-03 | 2021-01-08 | 深圳市彬讯科技有限公司 | Outbound processing method, device, equipment and medium for enterprise operation |
| CN112035624A (en) * | 2020-09-11 | 2020-12-04 | 上海明略人工智能(集团)有限公司 | Text recommendation method and device and storage medium |
| CN112330411A (en) * | 2020-11-17 | 2021-02-05 | 中国平安财产保险股份有限公司 | Group product recommendation method and device, computer equipment and storage medium |
| CN112330411B (en) * | 2020-11-17 | 2023-10-31 | 中国平安财产保险股份有限公司 | Group product recommendation method, group product recommendation device, computer equipment and storage medium |
| CN113763096A (en) * | 2020-12-14 | 2021-12-07 | 北京京东尚科信息技术有限公司 | Product shopping guide processing method, device, electronic equipment, medium and program product |
| CN112967141A (en) * | 2021-03-30 | 2021-06-15 | 建信金融科技有限责任公司 | On-line construction method, device, medium and electronic equipment for asset transaction |
| CN112884083A (en) * | 2021-03-31 | 2021-06-01 | 中国工商银行股份有限公司 | Intelligent outbound call processing method and device |
| CN113342964B (en) * | 2021-06-03 | 2022-04-19 | 云南大学 | Recommendation type determination method and system based on mobile service |
| CN113342964A (en) * | 2021-06-03 | 2021-09-03 | 云南大学 | Recommendation type determination method and system based on mobile service |
| WO2022267176A1 (en) * | 2021-06-23 | 2022-12-29 | 未鲲(上海)科技服务有限公司 | Product recommendation method and apparatus based on artificial intelligence, and device and storage medium |
| CN113468420A (en) * | 2021-06-29 | 2021-10-01 | 杭州摸象大数据科技有限公司 | Method and system for recommending products |
| CN113468420B (en) * | 2021-06-29 | 2024-04-05 | 杭州摸象大数据科技有限公司 | Product recommendation method and system |
| CN114049224A (en) * | 2021-10-13 | 2022-02-15 | 北京博瑞彤芸科技股份有限公司 | Training method and system for insurance businessman |
| CN114971795A (en) * | 2022-05-27 | 2022-08-30 | 中国平安财产保险股份有限公司 | Insurance product recommendation method and device, computer equipment and storage medium |
| CN117252995A (en) * | 2023-11-17 | 2023-12-19 | 深圳市加推科技有限公司 | Intelligent business card using method, intelligent sales system and storage medium |
| CN117252995B (en) * | 2023-11-17 | 2024-03-05 | 深圳市加推科技有限公司 | Intelligent business card using method, intelligent sales system and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110738545A (en) | Product recommendation method and device based on user intention identification, computer equipment and storage medium | |
| US11501763B2 (en) | Machine learning tool for navigating a dialogue flow | |
| CN110659318B (en) | Big data-based policy pushing method, system and computer equipment | |
| CN111079056A (en) | Method, device, computer equipment and storage medium for extracting user portrait | |
| US11804216B2 (en) | Generating training datasets for a supervised learning topic model from outputs of a discovery topic model | |
| CN112380331A (en) | Information pushing method and device | |
| US20220138770A1 (en) | Method and apparatus for analyzing sales conversation based on voice recognition | |
| CN111339284A (en) | Product intelligent matching method, device, equipment and readable storage medium | |
| CN110516057B (en) | Petition question answering method and device | |
| CN113112282A (en) | Method, device, equipment and medium for processing consult problem based on client portrait | |
| CN114266443A (en) | Data evaluation method and device, electronic equipment and storage medium | |
| JP2020140692A (en) | Sentence extracting system, sentence extracting method, and program | |
| CN112633690A (en) | Service personnel information distribution method, service personnel information distribution device, computer equipment and storage medium | |
| CN111914077A (en) | Customized speech recommendation method, device, computer equipment and storage medium | |
| CN112016850A (en) | Service evaluation method and device | |
| CN115481231A (en) | Session control method, device, computer equipment and storage medium | |
| CN113822071A (en) | Background information recommendation method and device, electronic equipment and medium | |
| US20210118433A1 (en) | Detecting extraneous topic information using artificial intelligence models | |
| CN111680497B (en) | Session recognition model training method and device | |
| CN114022165A (en) | Matching method, matching device, computer equipment and storage medium | |
| CN114239604A (en) | Online consultation processing method and device and computer equipment | |
| CN111882339A (en) | Prediction model training and response rate prediction method, device, equipment and storage medium | |
| CN117252995B (en) | Intelligent business card using method, intelligent sales system and storage medium | |
| CN112989785B (en) | Text vector acquisition method and device and text similarity calculation method and device | |
| KR102718884B1 (en) | Method and device for supporting b2b sales efforts using artificial intelligence model |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20200131 |
|
| WD01 | Invention patent application deemed withdrawn after publication |