CN110502639B

CN110502639B - Information recommendation method and device based on problem contribution degree and computer equipment

Info

Publication number: CN110502639B
Application number: CN201910615612.4A
Authority: CN
Inventors: 王伟
Original assignee: Ping An Bank Co Ltd
Current assignee: Ping An Bank Co Ltd
Priority date: 2019-07-09
Filing date: 2019-07-09
Publication date: 2022-09-06
Anticipated expiration: 2039-07-09
Also published as: CN110502639A

Abstract

The invention discloses an information recommendation method and device based on problem contribution degree and computer equipment. The method comprises the following steps: acquiring an initial associated problem list corresponding to each topic in the constructed initial knowledge graph; acquiring a first value problem chain set and a second value problem chain set according to the problem chains of all users in the selected target user community and corresponding operation result information; obtaining a problem retention degree queue ranked in ascending order according to the problem retention degrees of the problems in the second value problem chain set; acquiring a target question set of each topic; acquiring the number of the problems in the initial associated problem list, which is the same as that of the target problem set; and obtaining the questions with the problem contribution degree descending ranking not exceeding the target question replacing quantity in the problem contribution degree queue as a replacing question set. The method realizes the division of different user communities according to interests, the division of the historical access records into valuable and non-valuable problem chains, and the periodic calculation and adjustment of the problem contribution degree queue in the user communities.

Description

Information recommendation method and device based on problem contribution degree and computer equipment

Technical Field

The invention relates to the technical field of data analysis, in particular to an information recommendation method and device based on problem contribution degree, computer equipment and a storage medium.

Background

At present, with the rapid popularization of mobile internet, various enterprises have realized that users can purchase online products, such as financial products including financial products, structural deposit and currency funds, without going to a counter, and can directly complete the purchase operation on a mobile phone through an APP (APP is an application program for short). However, the existing APP has the defects of multiple financial product purchasing processes, complex display interface, long time for the user to become familiar with the product, low accuracy in identifying the intention of the user and low customer conversion rate.

Disclosure of Invention

The embodiment of the invention provides an information recommendation method, an information recommendation device, computer equipment and a storage medium based on problem contribution degree, and aims to solve the problems that when a user recommends a product on line in the prior art, the user knows many product processes, the display interface is complex, the user needs to spend a long time to be familiar with the product, the recognition accuracy rate of the user intention is not high, and the customer conversion rate is reduced.

In a first aspect, an embodiment of the present invention provides an information recommendation method based on a problem contribution degree, including:

the method comprises the steps of constructing an initial knowledge graph of a product purchasing process corresponding to a plurality of target user communities one by one in advance;

acquiring an initial associated problem list corresponding to each topic in the initial knowledge graph;

according to the problem chains of all users in the selected target user community and operation result information corresponding to all problems, acquiring a first value problem chain set used for representing valuable problems and a second value problem chain set used for representing non-valuable problems;

calculating and obtaining the question retention degree of each question in the second-value question chain set to obtain a question retention degree queue which corresponds to each topic and is ranked according to the question retention degree of the question in an ascending order;

obtaining the questions with the ranking of the question retention degree of each question in the question retention degree queue before a preset ranking threshold value, and using the obtained questions as a target question set of each topic;

acquiring the number of the problems in the initial associated problem list, which is the same as that of the target problem set, and taking the number of the problems as the target problem replacement number;

obtaining the problems of which the descending rank of the contribution degree of the problems in the problem contribution degree queue corresponding to each topic in the initial knowledge graph does not exceed the target problem replacement number to serve as the replacement problem set of each topic; and

and replacing the same questions in the initial associated question list and the target question set by corresponding replacement question sets to obtain an updated associated question list of each topic.

In a second aspect, an embodiment of the present invention provides an information recommendation apparatus based on a problem contribution degree, including:

the system comprises a knowledge graph construction unit and a knowledge graph analysis unit, wherein the knowledge graph construction unit is used for constructing initial knowledge graphs of product purchase processes corresponding to a plurality of target user communities one by one in advance;

an initial question list acquiring unit, configured to acquire an initial associated question list corresponding to each topic in the initial knowledge graph;

the problem chain set dividing unit is used for acquiring a first price problem chain set used for representing a valuable problem and a second price problem chain set used for representing a non-valuable problem according to the problem chains of all the users in the selected target user community;

the question retention degree calculating unit is used for calculating and acquiring the question retention degree of each question in the second-value question chain set so as to obtain a question retention degree queue which corresponds to each topic and is ranked according to the question retention degree of the question in an ascending order;

a target problem set obtaining unit, configured to obtain a problem in the problem retention degree queue, where a rank of the problem retention degree of each problem is before a preset rank threshold, so as to serve as a target problem set of each topic;

a replacement quantity acquiring unit, configured to acquire the same number of questions as the target question set in the initial associated question list, as a target question replacement quantity;

a replacement problem set acquisition unit, configured to acquire problems in which the descending order ranking of the problem contribution degrees in the problem contribution degree queue corresponding to each topic in the initial knowledge graph does not exceed the target problem replacement number, so as to serve as a replacement problem set for each topic; and

and the question replacing unit is used for replacing the same question in the initial associated question list and the target question set by a corresponding replacing question set to obtain an updated associated question list of each topic.

In a third aspect, an embodiment of the present invention further provides a computer device, which includes a memory, a processor, and a computer program stored on the memory and executable on the processor, where the processor, when executing the computer program, implements the information recommendation method based on the problem contribution degree according to the first aspect.

In a fourth aspect, the embodiment of the present invention further provides a computer-readable storage medium, where the computer-readable storage medium stores a computer program, and the computer program, when executed by a processor, causes the processor to execute the information recommendation method based on the problem contribution degree according to the first aspect.

The embodiment of the invention provides an information recommendation method and device based on problem contribution degree, computer equipment and a storage medium. The method comprises the steps of constructing an initial knowledge graph of a product purchasing process corresponding to a plurality of target user communities one by one in advance; acquiring an initial associated problem list corresponding to each topic in the initial knowledge graph; acquiring a first price problem chain set used for representing a valuable problem and a second price problem chain set used for representing a non-valuable problem according to the problem chain of each user in the selected target user community and operation result information corresponding to each problem; calculating and obtaining the question retention degree of each question in the second-value question chain set to obtain a question retention degree queue which corresponds to each topic and is ranked according to the question retention degree of the question in an ascending order; obtaining the questions with the ranking of the question retention degree of each question in the question retention degree queue before a preset ranking threshold value, and using the obtained questions as a target question set of each topic; acquiring the number of the problems in the initial associated problem list, which is the same as that of the target problem set, and taking the number of the problems as the target problem replacement number; obtaining the problems of which the descending rank of the contribution degree of the problems in the problem contribution degree queue corresponding to each topic in the initial knowledge graph does not exceed the target problem replacement number to serve as the replacement problem set of each topic; and replacing the same questions in the initial associated question list and the target question set by corresponding replacement question sets to obtain an updated associated question list of each topic. The method realizes accurate recognition of scene-topic attribution degree of the problem by adopting a deep learning model, constructs a related problem set, divides different user communities according to interests, divides historical access records into valuable problem chains and non-valuable problem chains, periodically calculates and adjusts a problem contribution degree queue in the user communities, and improves the recommendation success rate.

Drawings

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

Fig. 1 is a schematic view of an application scenario of an information recommendation method based on a problem contribution degree according to an embodiment of the present invention;

fig. 2 is a schematic flowchart of an information recommendation method based on problem contribution according to an embodiment of the present invention;

fig. 3 is a schematic diagram of an initial knowledge graph in the problem contribution-based information recommendation method according to the embodiment of the present invention;

fig. 4 is a sub-flow diagram of an information recommendation method based on problem contribution according to an embodiment of the present invention;

fig. 5 is another sub-flow diagram of an information recommendation method based on problem contribution according to an embodiment of the present invention;

fig. 6 is a schematic diagram of first value problem chains included in a first value problem chain set in the problem contribution-based information recommendation method according to the embodiment of the present invention;

fig. 7 is a schematic diagram of second value question chains included in a second value question chain set in the question contribution-based information recommendation method according to the embodiment of the present invention;

FIG. 8 is a schematic block diagram of an information recommendation apparatus based on problem contribution according to an embodiment of the present invention;

FIG. 9 is a block diagram schematically illustrating sub-units of an information recommendation apparatus based on problem contribution according to an embodiment of the present invention;

FIG. 10 is a schematic block diagram of another sub-unit of an information recommendation device based on question contribution degree according to an embodiment of the present invention;

FIG. 11 is a schematic block diagram of a computer device provided by an embodiment of the present invention.

Detailed Description

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

It will be understood that the terms "comprises" and/or "comprising," when used in this specification and the appended claims, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.

It is also to be understood that the terminology used in the description of the invention herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. As used in the specification of the present invention and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise.

It should be further understood that the term "and/or" as used in this specification and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

Referring to fig. 1 and fig. 2, fig. 1 is a schematic view of an application scenario of an information recommendation method based on a problem contribution degree according to an embodiment of the present invention, and fig. 2 is a schematic view of a flow of the information recommendation method based on a problem contribution degree according to an embodiment of the present invention.

As shown in fig. 2, the method includes steps S110 to S180.

S110, constructing an initial knowledge graph of a product purchasing process corresponding to a plurality of target user communities one by one in advance.

In this embodiment, when the user accesses the UI interface (i.e., the user interaction interface) for online product recommendation provided by the server, an initial knowledge graph of the product purchasing process corresponding to a plurality of target user communities one by one needs to be constructed in the server in advance.

In the application, the knowledge graph comprises a conversation scene and a topic, and the one-time purchasing behavior of a user is divided into a plurality of indispensable general links to form the conversation scene.

Taking a financial product purchasing behavior as an example, the purchasing process can be divided into the following scenes: product consultation, transaction flow consultation, transaction income consultation and after-sale consultation.

Each dialogue scene comprises a plurality of topics, the granularity of topic expression is smaller than that of the dialogue scene, and the topic expression is used for further dividing and confirming the real intention of the client.

As shown in fig. 3, in the product consultation scenario, the following are included: product definition, product characteristics, product subscription time, product income, product sale and the like. All topics constitute all the user's intentions for a scene. Each topic contains a number of questions, which are a set of questions that the user may pose in order to learn the connotation of each topic. Taking the product definition as an example, the following problems may be included: 1) what this product is; 2) what this product defines is.

Through the pre-constructed knowledge graph, common scenes and common problems in the process of purchasing products by a user are summarized effectively.

And S120, acquiring an initial associated problem list corresponding to each topic in the initial knowledge graph.

In this embodiment, in order to obtain the initial associated question list corresponding to each topic in the initial knowledge graph, each question Q needs to be defined first _i Topic attribution degree of

This index is used to calculate the order in which each question is displayed in the final initial list of associated questions. Further, in the present invention,

the index is again attributed to the scene

Degree of topic attribution

The product of (a) constitutes, namely:

because each scene and topic is a phrase, firstly calculating the semantic similarity between each question and the scene; and then calculating the semantic similarity between each question and each topic.

In an embodiment, step S110 further includes:

extracting keywords from each sample problem in the input sample problem set through a word frequency-inverse text frequency index model to obtain keywords corresponding to each sample problem;

converting the keywords respectively corresponding to each sample question into corresponding word vectors through a conversion model for converting the keywords into the word vectors;

carrying out scene type marking and topic type marking on each sample question to obtain a scene attribution degree and a topic attribution degree corresponding to each sample question;

and taking the word vector corresponding to each sample question as the input of the deep convolutional neural network to be trained, taking the column vector consisting of the scene attribution degree and the topic attribution degree corresponding to each sample question as the output of searching the deep convolutional neural network to be trained, and training the deep convolutional neural network to be trained to obtain a deep convolutional neural network model for judging the topic attribution of the question.

Namely, the semantic similarity calculation adopts a deep learning model, and the steps are as follows:

A1) firstly, a certain number of sample problems are collected and converted into doc2vec vectors, and the vectors are labeled according to the type of a conversational scene and the type of a conversational question and divided into a training set and a testing set.

A2) And training and verifying by using a deep convolutional neural network, and calculating the probability of belonging to each scene type and topic type, namely the scene attribution degree and the topic attribution degree.

And after calculating all the scene-topic attribution degrees of each question, dividing the question into topics with the highest scene-topic attribution degree values.

In one embodiment, step S120 includes:

acquiring a question list corresponding to each topic in the initial knowledge graph;

calculating and obtaining the similarity between each question corresponding to each question list in the initial knowledge graph and the corresponding topic, and obtaining a target question set of which the descending rank of the similarity between each question and the corresponding topic is positioned before a preset ranking threshold value so as to form an initial associated question list corresponding to each topic.

After obtaining the associated question list with each topic in the initial knowledge graph, a certain number of questions are prepared in advance for each topic. Due to the fact that the display interface of the APP is limited, for the questions belonging to a certain topic, a related question list, namely the first K questions, is displayed according to the scene-topic attribution degree score from high to low. (K general experience is 3, 4, 5).

More specifically, upon access to a list of associated questions on the same topic, each topic is displayed with a label at the bottom of the APP. When a user clicks a label of a topic for the first time, firstly displaying an associated question list; next, each time one question in the list of associated questions is clicked, the answer to that question will be displayed in the APP, and the remaining ones of the K questions that were not accessed will be displayed in the area below the answer. For accesses accessing different topics, clicking different tags in the APP can switch topics back and forth.

And S130, acquiring a first price problem chain set for representing valuable problems and a second price problem chain set for representing non-valuable problems according to the problem chains of the users in the selected target user community and operation result information corresponding to the problems.

In this embodiment, online products such as financial products are clustered according to characteristics such as price, profitability, risk, and redemption period, and since there are many types of clustering methods, including K-Means, K-center point, etc., the application does not limit the specific clustering method. The financial products are classified into different categories by clustering. And dividing users who purchase the same type of products into the same user community.

In the present application, a user community that selects a product a to purchase is described as an example. The user of the user community who purchased product a asks a series of questions to reach the final result (i.e., purchase or unpurcure product a) in the process of asking the product, and the series of questions asked by each user for product a forms a question chain for the user. Some of these problem chains can be considered as first-value problem chains, and the other can be considered as second-value problem chains, and the specific division is defined by reference to the following definitions:

each first value question chain in the set of first value question chains can be understood as a value question chain, which is defined as a question chain that is visited by the user to promote the sale of a product, i.e., all questions visited in the whole flow from the first question of the user's login click on the first scenario to the last completion of the product purchase, as shown in fig. 6.

Each second value question chain in the set of second value question chains can be understood as a worthless question chain, which is defined as a question chain that does not contribute to the sale of a product, and all questions visited in the whole process from the first question of the first scenario clicked by the user to the last exit without purchasing the product, as shown in fig. 7. In practice, setting a time-break threshold value, such as 1 minute, 2 minutes, may be considered. Exceeding this threshold may be considered to cancel a purchase.

In one embodiment, as shown in fig. 4, as a first embodiment of step S130, step S130 includes:

s1311, obtaining operation result information corresponding to the problem chains of the users in the selected target user community;

s1312, if operation result information corresponding to the problem chain of the user is a successful trigger instruction, dividing the problem chain of the corresponding user into the first value problem chain set;

and S1313, if the operation result information corresponding to the problem chain of the user is an unsuccessful triggering instruction, dividing the problem chain corresponding to the user into the second value problem chain set.

In the first embodiment of step S130, whether the question chain of each user is divided into the first price question chain set or the second price question chain set is determined by whether the product was successfully purchased. In this way, each problem can be accurately classified.

In one embodiment, as shown in fig. 5, as a second embodiment of step S130, step S130 includes:

s1321, obtaining operation result information corresponding to the problem chain of each user in the selected target user community;

s1322, if the operation result information corresponding to the problem chain of the user is a successful trigger instruction, dividing the problem chain of the corresponding user into the first value problem chain set;

s1323, if the operation result information corresponding to the problem chain of the user is an unsuccessful triggering instruction, dividing the problem chain of the corresponding user into the second value problem chain set;

and S1324, if the reply time interval of each question corresponding to the question chain of the user exceeds a preset time threshold, dividing the question chain of the corresponding user into the second-value question chain set.

In the second embodiment of step S130, the same as the first embodiment is that whether the product is successfully purchased is used to determine that the problem chains of each user are divided into the first value problem chain set, but the difference is that the problem chains divided into the second value problem chain set are not successful trigger instructions except for the operation result information corresponding to the problem chains of the user, and the problem chains corresponding to the user are also divided into the second value problem chain set when the reply time interval of the problem chains of the user corresponding to each problem exceeds the preset time threshold (which indicates that the user is hesitant in the purchase process and the purchase will not be strong). In this way, each problem can be accurately classified.

In an embodiment, step S130 is followed by:

obtain the firstCollecting the total contribution degree of each problem to a preset statistical parameter by a one-price problem chain; wherein, the total contribution degree of each question in the first value question chain set to the statistical parameter is marked as C and

the total number of the corresponding users and the problem chains in the first value problem chain set is n, and the total contribution degree of the problem chain corresponding to the user i in the first value problem chain set to the statistical parameter is marked as C _i And is

The total number of the problems of the problem chain corresponding to the user i in the first value problem chain set is m, and the contribution degree of the problem k corresponding to the problem chain corresponding to the user i is marked as C _ik The contribution coefficient of the problem k pair in the problem chain corresponding to the user i is recorded as

And is provided with

The statistical parameter corresponding to the problem chain corresponding to the user i is recorded as sum _i ；

And according to the topics to which the questions belong in the first value question chain set, obtaining a question contribution degree queue which corresponds to the topics and is ranked according to the total contribution degree of the questions in a descending order.

In the present embodiment, every other period T, the contribution C of each question in the existing "valuable question chain" to the sales of the final product is calculated for a particular community of users. Let Ci be the contribution of the question i to the final product sales, and the contribution C is made up of the product of the contribution coefficient and the actual sales of the product realized by the value question chain. The specific process is as follows:

the actual sales of the product achieved by the valuable problem chain is defined as sae _ sum. If the valuable question chain of the user i contains m questions, and the question sequence is from back to front, the contribution coefficient of the kth question in the valuable question chain

Comprises the following steps:

through the process, the contribution degree of each question in the valuable question chain of the user i to the final product sales can be calculated, the total contribution degree of each question in a user community is calculated, all the questions are judged to belong to the topics, and then the questions are sorted according to the total contribution degree of the questions under the topics to form a question contribution degree queue corresponding to each topic.

S140, calculating and obtaining the question retention degree of each question in the second-value question chain set to obtain a question retention degree queue which corresponds to each topic and is ranked according to the question retention degree of the question in an ascending order.

In the present embodiment, by

Calculating the question retention degree of each question in the second value question chain set; wherein, the J-th question in the question chain corresponding to the user I in the second value question chain set is marked as Keeprate _I And the problem chain corresponding to the user I in the second value problem chain set comprises N problems. In a period, recording all questions visited by all users in the same user community each time from login to logout, so that X questions with the minimum question retention degree in each topic can be found. In practice, the value of X needs to be adjusted continuously.

And S150, obtaining the questions with the ranking of the question retention degree of each question in the question retention degree queue before a preset ranking threshold value, and using the obtained questions as the target question set of each topic.

In this embodiment, in order to replace an unproductive question in the initial associated question list, a question whose rank of the question retention of each question in the question retention queue is before a preset ranking threshold may be acquired as a target question set of each topic. These problems in the set of target problems are less effective in facilitating the purchase of a product by the user and may be replaced by problems that are more effective in facilitating the purchase of a product by the user.

And S160, acquiring the number of the problems in the initial associated problem list, which is the same as that of the target problem set, and taking the number of the problems as the target problem replacement number.

In this embodiment, in order to replace an unproductive question in the initial associated question list, the same number of questions as the target question set in the initial associated question list needs to be obtained. The same problems exist in the initial associated problem list and the target problem set, which means that the same problems can be replaced, and the replaced problems need to be obtained from a problem contribution degree queue, so that accurate replacement of low contribution degree problems is realized.

S170, obtaining the questions with the contribution degree descending ranking not exceeding the target question replacing quantity in the question contribution degree queue corresponding to each topic in the initial knowledge graph to serve as replacing question sets of each topic.

In this embodiment, when a problem with a high contribution degree in the problem contribution degree queue is obtained, a problem with a ranking that does not exceed the target problem replacement number needs to be extracted from the problem contribution degree queue, and these top-ranked problems can further promote understanding and selling of products, and realize accurate replacement of a problem with a low contribution degree.

And S180, replacing the same questions in the initial associated question list and the target question set through corresponding replacement question sets to obtain an updated associated question list of each topic.

In the embodiment, namely in each topic, the question with the contribution degree of the question in the topic, which is ranked at the top X bits in the queue, is used to replace the X questions with the minimum "question retention degree". The next time a user of the community accesses a topic, a new list of associated questions will be accessed.

The method realizes accurate recognition of scene-topic attribution degree of the problem by adopting a deep learning model, constructs a related problem set, divides different user communities according to interests, divides historical access records into valuable problem chains and non-valuable problem chains, periodically calculates and adjusts a problem contribution degree queue in the user communities, and improves the recommendation success rate.

The embodiment of the invention also provides an information recommendation device based on the problem contribution degree, which is used for executing any embodiment of the information recommendation method based on the problem contribution degree. Specifically, referring to fig. 8, fig. 8 is a schematic block diagram of an information recommendation device based on problem contribution according to an embodiment of the present invention. The information recommendation device 100 based on the problem contribution degree may be configured in a server.

As shown in fig. 8, the information recommendation device 100 based on the question contribution degree includes a knowledge graph construction unit 110, an initial question list acquisition unit 120, a question chain set division unit 130, a question retention degree calculation unit 140, a target question set acquisition unit 150, a replacement number acquisition unit 160, a replacement question set acquisition unit 170, and a question replacement unit 180.

The knowledge graph constructing unit 110 is configured to construct an initial knowledge graph of a product purchasing process in one-to-one correspondence with a plurality of target user communities in advance.

Each dialogue scene also comprises a plurality of topics, the granularity of topic expression is smaller than that of the dialogue scene, and the topic expression is further used for dividing and confirming the true intention of the client.

As shown in fig. 3, in the product consultation scenario, the following are included: product definition, product characteristics, product subscription time, product income, product sale and the like. All topics constitute all the user's intentions for a scene. Each topic, in turn, contains a number of questions, which are a set of questions that the user may pose in order to learn the connotation of each topic. Taking the product definition as an example, the following problems may be included: 1) what this product is; 2) what this product defines is.

An initial question list acquiring unit 120, configured to acquire an initial associated question list corresponding to each topic in the initial knowledge graph.

This index is used to calculate the order in which each question is displayed in the final initial list of associated questions. Further, in the present invention, it is preferable that,

the index is again attributed to the scene

Degree of topic affiliation

The product of (a) constitutes, i.e.:

In one embodiment, the information recommendation device based on the problem contribution degree further includes:

the keyword extraction unit is used for extracting keywords from each sample question in the input sample question set through a word frequency-inverse text frequency index model to obtain keywords corresponding to each sample question;

the word vector conversion unit is used for converting the keywords respectively corresponding to each sample question into corresponding word vectors through a conversion model for converting the keywords into the word vectors;

the attribution degree calculating unit is used for carrying out scene type marking and topic type marking on each sample question to obtain scene attribution degree and topic attribution degree corresponding to each sample question;

and the model training unit is used for taking the word vector corresponding to each sample question as the input of the deep convolutional neural network to be trained, taking the column vector consisting of the scene attribution degree and the topic attribution degree corresponding to each sample question as the output for searching the deep convolutional neural network to be trained, and training the deep convolutional neural network to be trained to obtain the deep convolutional neural network model for judging the topic attribution of the question.

In one embodiment, the initial problem list obtaining unit includes:

a question list acquiring unit for acquiring a question list corresponding to each topic in the initial knowledge graph;

and the question screening unit is used for calculating and acquiring the similarity between each question corresponding to each question list in the initial knowledge graph and the corresponding topic, and acquiring a target question set of which the descending rank of the similarity between each question and the corresponding topic is positioned before a preset rank threshold value so as to form an initial associated question list corresponding to each topic.

After obtaining the associated question list with each topic in the initial knowledge graph, a certain number of questions are prepared in advance for each topic. Due to the fact that the APP display interface is limited, for the questions belonging to a certain topic, a related question list, namely the first K questions, is displayed according to the scene-topic attribution degree score from high to low. (general experience of K is 3, 4, 5).

More specifically, when visiting a list of associated questions on the same topic, each topic is displayed with a label at the bottom of the APP. When a user clicks a label of a topic for the first time, firstly displaying an associated question list; next, each time one question in the list of associated questions is clicked, the answer to that question will be displayed in the APP, and the remaining ones of the K questions that were not accessed will be displayed in the area below the answer. For accesses accessing different topics, clicking different tags in the APP can switch the topics back and forth.

The problem chain set dividing unit 130 is configured to obtain a first value problem chain set representing a valuable problem and a second value problem chain set representing a non-valuable problem according to the problem chain of each user in the selected target user community and operation result information corresponding to each problem.

In the present application, a user community that selects a product a to purchase is described as an example. The user of the user community who purchased product a asks a series of questions to reach the final result (i.e., purchase or unpurcure product a) in the process of asking the product, and the series of questions asked by each user for product a forms a question chain for the user. Some of these problem chains can be regarded as a first-value problem chain, and the other part can be regarded as a second-value problem chain, and the specific division is defined by reference to the following:

In an embodiment, as shown in fig. 9, as a first embodiment of the question chain set dividing unit 130, the question chain set dividing unit 130 includes:

a first operation result information obtaining unit 1311 configured to obtain operation result information corresponding to a problem chain of each user in the selected target user community;

a first dividing unit 1312, configured to divide the problem chain corresponding to the user into the first value problem chain set if the operation result information corresponding to the problem chain of the user is a successful trigger instruction;

a second dividing unit 1313, configured to divide the problem chain corresponding to the user into the second value problem chain set if the operation result information corresponding to the problem chain of the user is an unsuccessful trigger instruction.

In the first embodiment of the question chain set dividing unit 130, whether the question chain of each user is divided into the first value question chain set or the second value question chain set is determined by whether the product was successfully purchased. In this way, each problem can be accurately classified.

In an embodiment, as shown in fig. 10, as a second embodiment of the problem chain set dividing unit 130, the problem chain set dividing unit 130 includes:

a second operation result information obtaining unit 1321 configured to obtain operation result information corresponding to the problem chain of each user in the selected target user community;

a third dividing unit 1322, configured to divide the problem chain corresponding to the user into the first value problem chain set if the operation result information corresponding to the problem chain of the user is a successful trigger instruction;

a fourth dividing unit 1323, configured to divide the problem chain corresponding to the user into the second value problem chain set if the operation result information corresponding to the problem chain of the user is an unsuccessful trigger instruction;

a fifth dividing unit 1324, configured to divide the question chain of the corresponding user into the second value question chain set if the response time interval of the question chain of the corresponding user to each question exceeds a preset time threshold.

In the second embodiment of the step problem chain set dividing unit 130, the same as the first embodiment is to determine whether the problem chain of each user is divided into the first value problem chain set by whether the product is successfully purchased, but the difference is that the problem chain divided into the second value problem chain set is not the successful trigger instruction except for the operation result information corresponding to the problem chain of the user, and the problem chain of the corresponding user is also divided into the second value problem chain set when the reply time interval of the problem chain of the user corresponding to each problem exceeds the preset time threshold (indicating that the user is hesitant in the purchasing process and the purchasing will not be strong). In this way, each problem can be accurately classified.

In one embodiment, the information recommendation device based on question contribution further includes:

a total contribution degree obtaining unit, configured to obtain a total contribution degree of each problem of the first value problem chain set to a preset statistical parameter; wherein, the total contribution degree of each question in the first value question chain set to the statistical parameter is marked as C and

the total number of the corresponding users and the problem chains in the first value problem chain set is n, and the total contribution degree of the problem chains corresponding to the users i in the first value problem chain set to the statistical parameters is marked as C _i And is provided with

The total number of the problems of the problem chain corresponding to the user i in the first value problem chain set is m, and the contribution degree of the problem k in the problem chain corresponding to the user i to the problem chain is marked as C _ik The contribution coefficient of the problem k pair in the problem chain corresponding to the user i is recorded as

And is

And the question contribution degree queue acquisition unit is used for acquiring a question contribution degree queue which corresponds to each topic and is ranked according to the total contribution degree of the questions in a descending order according to the topics to which the questions belong in the first value question chain set.

the actual sales of the product achieved by the valuable problem chain is defined as sae _ sum. If the valuable question chain of the user i contains m questions, and the question sequence is from back to front, the contribution coefficient of the kth question in the valuable question chain is set

Comprises the following steps:

And the question retention degree calculating unit 140 is used for calculating and acquiring the question retention degree of each question in the second value question chain set so as to obtain a question retention degree queue which corresponds to each topic and is ranked according to the question retention degree of the question in an ascending order.

In the present embodiment, by

Calculating the question retention of each question in the second value question chain set; wherein, the J-th question in the question chain corresponding to the user I in the second value question chain set is marked as KeepRate _I And the problem chain corresponding to the user I in the second value problem chain set comprises N problems. In a period, recording all questions visited by all users in the same user community each time from login to logout, so that X questions with the minimum question retention degree in each topic can be found. In practice, the value of X needs to be adjusted continuously.

A target question set acquiring unit 150, configured to acquire, as a target question set for each topic, a question with a rank of the question retention of each question in the question retention queue before a preset ranking threshold.

A replacement quantity obtaining unit 160, configured to obtain the same number of questions as the target question set in the initial associated question list as a target question replacement quantity.

A replacement problem set obtaining unit 170, configured to obtain, as a replacement problem set for each topic, the problems in the problem contribution degree queue corresponding to each topic in the initial knowledge graph, in which the descending order of the problem contribution degrees is not beyond the target problem replacement number.

In this embodiment, when a problem with a high contribution degree in the problem contribution degree queue is obtained, a problem with a rank that does not exceed the replacement number of the target problem needs to be extracted from the problem contribution degree queue, and the top-ranked problems can further promote understanding and sales of products, so as to realize accurate replacement of a problem with a low contribution degree.

A question replacing unit 180, configured to replace the same question in the initial associated question list as the target question set with a corresponding replaced question set, so as to obtain an updated associated question list for each topic.

The device realizes accurate recognition of scene-topic attribution degree of the problem by adopting a deep learning model, constructs a related problem set, divides different user communities according to interests, divides historical access records into valuable problem chains and non-valuable problem chains, periodically calculates and adjusts a problem contribution degree queue in the user communities, and improves the recommendation success rate.

The information recommendation apparatus based on the problem contribution degree may be implemented in the form of a computer program that can be run on a computer device as shown in fig. 11.

Referring to fig. 11, fig. 11 is a schematic block diagram of a computer device according to an embodiment of the present invention. The computer device 500 is a server, and the server may be an independent server or a server cluster composed of a plurality of servers.

Referring to fig. 11, the computer device 500 includes a processor 502, memory, and a network interface 505 connected by a system bus 501, where the memory may include a non-volatile storage medium 503 and an internal memory 504.

The non-volatile storage medium 503 may store an operating system 5031 and a computer program 5032. The computer program 5032, when executed, may cause the processor 502 to perform an information recommendation method based on the problem contribution degree.

The processor 502 is used to provide computing and control capabilities that support the operation of the overall computer device 500.

The internal memory 504 provides an environment for the operation of the computer program 5032 in the non-volatile storage medium 503, and when the computer program 5032 is executed by the processor 502, the processor 502 can be caused to execute the information recommendation method based on the problem contribution degree.

The network interface 505 is used for network communication, such as providing transmission of data information. Those skilled in the art will appreciate that the configuration shown in fig. 11 is a block diagram of only a portion of the configuration associated with aspects of the present invention and is not intended to limit the computing device 500 to which aspects of the present invention may be applied, and that a particular computing device 500 may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.

The processor 502 is configured to run the computer program 5032 stored in the memory to implement the information recommendation method based on the problem contribution degree according to the embodiment of the present invention.

Those skilled in the art will appreciate that the embodiment of a computer device illustrated in fig. 11 does not constitute a limitation on the specific construction of the computer device, and that in other embodiments a computer device may include more or fewer components than those illustrated, or some components may be combined, or a different arrangement of components. For example, in some embodiments, the computer device may only include a memory and a processor, and in such embodiments, the structures and functions of the memory and the processor are consistent with those of the embodiment shown in fig. 11, and are not described herein again.

It should be understood that, in the embodiment of the present invention, the Processor 502 may be a Central Processing Unit (CPU), and the Processor 502 may also be other general purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field Programmable Gate Arrays (FPGAs) or other Programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Wherein a general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.

In another embodiment of the present invention, a computer-readable storage medium is provided. The computer readable storage medium may be a non-volatile computer readable storage medium. The computer readable storage medium stores a computer program, wherein the computer program, when executed by a processor, implements the information recommendation method based on the problem contribution degree according to the embodiment of the present invention.

It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described apparatuses, devices and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again. Those of ordinary skill in the art will appreciate that the elements and algorithm steps of the examples described in connection with the embodiments disclosed herein may be embodied in electronic hardware, computer software, or combinations of both, and that the components and steps of the examples have been described in a functional general in the foregoing description for the purpose of illustrating clearly the interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.

In the embodiments provided by the present invention, it should be understood that the disclosed apparatus, device and method can be implemented in other ways. For example, the above-described embodiments of the apparatus are merely illustrative, and for example, the division of the units is only a logical division, and there may be other divisions when the actual implementation is performed, or units having the same function may be grouped into one unit, for example, a plurality of units or components may be combined or may be integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may also be an electric, mechanical or other form of connection.

The units described as separate parts may or may not be physically separate, and parts displayed as units 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 units can be selected according to actual needs to achieve the purpose of the solution of the embodiment of the present invention.

In addition, functional units in the embodiments of the present invention 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, and can also be realized in a form of a software functional unit.

The integrated unit, if implemented in the form of a software functional unit and sold or used as a separate product, may be stored in a storage medium. Based on such understanding, the technical solution of the present invention essentially or partly contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a Read-Only Memory (ROM), a magnetic disk, or an optical disk.

While the invention has been described with reference to specific embodiments, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. An information recommendation method based on problem contribution degree is characterized by comprising the following steps:

acquiring a first price problem chain set used for representing a valuable problem and a second price problem chain set used for representing a non-valuable problem according to the problem chain of each user in the selected target user community and operation result information corresponding to each problem;

obtaining the questions with the contribution degree descending ranking not exceeding the target question replacing number in the question contribution degree queue corresponding to each topic in the initial knowledge graph as a replacing question set of each topic; and

replacing the same questions in the initial associated question list and the target question set by corresponding replacement question sets to obtain updated associated question lists of all topics;

after acquiring a first value problem chain set for representing a valuable problem and a second value problem chain set for representing a non-valuable problem according to the problem chains of the users in the selected target user community, the method further comprises the following steps:

acquiring the total contribution of each question of the first price question chain set to a preset statistical parameter; wherein, the total contribution degree of each question in the first value question chain set to the statistical parameter is marked as C and

The total number of questions in the question chain corresponding to the user i in the first value question chain set is m, and the contribution degree of the question k in the question chain corresponding to the user i to the question chain is recorded as m

The contribution coefficient of the problem k pair in the problem chain corresponding to the user i is recorded as

And is

And recording the statistical parameters corresponding to the problem chain corresponding to the user i

；

According to the topics to which the questions belong in the first value question chain set, obtaining question contribution degree queues which correspond to the topics and are ranked according to the total contribution degrees of the questions in a descending order;

the step of obtaining the question retention degree of each question in the second value question chain set through calculation to obtain a question retention degree queue which corresponds to each topic and is ranked according to the question retention degree of the question in an ascending order, comprises the following steps:

by passing

Calculating the question retention of each question in the second value question chain set; wherein, the J-th question in the question chain corresponding to the user I in the second value question chain set is marked as the question chain

And the problem chain corresponding to the user I in the second value problem chain set comprises N problems.

2. The information recommendation method based on question contribution degree according to claim 1, wherein before the pre-constructing the initial knowledge graph of the product purchase process corresponding to a plurality of target user communities, further comprising:

extracting keywords of each sample problem in the input sample problem set through a word frequency-inverse text frequency index model to obtain keywords corresponding to each sample problem;

carrying out scene type labeling and topic type labeling on each sample question to obtain a scene attribution degree and a topic attribution degree corresponding to each sample question;

3. The method of claim 2, wherein the obtaining of an initial list of associated questions corresponding to each topic in the initial knowledge graph comprises:

and calculating and obtaining the similarity between each question corresponding to each question list in the initial knowledge graph and the corresponding topic, and obtaining a target question set of which the descending rank of the similarity between each question and the corresponding topic is positioned before a preset ranking threshold value so as to form an initial associated question list corresponding to each topic.

4. The method of claim 1, wherein the obtaining a first value problem chain set representing valuable problems and a second value problem chain set representing worthless problems according to the problem chains of the users in the selected target user community and operation result information corresponding to the problems comprises:

obtaining operation result information corresponding to the problem chain of each user in the selected target user community;

if the operation result information corresponding to the problem chain of the user is a successful trigger instruction, dividing the problem chain of the corresponding user into the first value problem chain set;

and if the operation result information corresponding to the problem chain of the user is the unsuccessful triggering instruction, dividing the problem chain corresponding to the user into the second value problem chain set.

5. The method as claimed in claim 1, wherein the obtaining a first value problem chain set representing a valuable problem and a second value problem chain set representing a non-valuable problem according to the problem chains of the users in the selected target user community and the operation result information corresponding to the problems comprises:

acquiring operation result information corresponding to the problem chain of each user in the selected target user community;

if the operation result information corresponding to the problem chain of the user is an unsuccessful triggering instruction, dividing the problem chain of the corresponding user into the second value problem chain set;

and if the reply time interval of the problem chain of the user corresponding to each problem exceeds a preset time threshold, dividing the problem chain of the corresponding user into the second-value problem chain set.

6. An information recommendation apparatus based on a problem contribution degree, comprising:

the system comprises a knowledge graph construction unit, a product purchase unit and a product purchase unit, wherein the knowledge graph construction unit is used for constructing initial knowledge graphs of product purchase processes corresponding to a plurality of target user communities one by one in advance;

a problem chain set dividing unit, which is used for acquiring a first value problem chain set used for representing a valuable problem and a second value problem chain set used for representing a non-valuable problem according to the problem chain of each user in the selected target user community;

the question retention degree calculating unit is used for calculating and obtaining the question retention degree of each question in the second value question chain set so as to obtain a question retention degree queue which corresponds to each topic and is ranked according to the question retention degree of the question in an ascending order;

a target question set acquiring unit, configured to acquire a question with a question retention degree ranking of each question in the question retention degree queue before a preset ranking threshold, as a target question set of each topic;

a replacement question set obtaining unit, configured to obtain questions whose contribution degrees in a question contribution degree queue corresponding to each topic in the initial knowledge graph are ranked in a descending order without exceeding the target question replacement number, so as to serve as a replacement question set for each topic; and

a question replacing unit, configured to replace the same question in the initial associated question list as the target question set with a corresponding replacement question set, so as to obtain an updated associated question list of each topic;

the information recommendation device based on the problem contribution further comprises:

Question of question chain corresponding to user i in first value question chain setThe total number of questions is m, and the contribution degree of the question k in the question chain corresponding to the user i is recorded as

And is

；

A question contribution degree queue obtaining unit, configured to obtain, according to the topics to which the questions belong in the first value question chain set, question contribution degree queues that correspond to the topics and are ranked in a descending order according to total contribution degrees of the questions;

the problem retention degree calculation unit is specifically configured to:

by passing

Calculating the question retention of each question in the second value question chain set; wherein, the J-th question in the question chain corresponding to the user I in the second value question chain set is marked as

7. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the information recommendation method based on the problem contribution degree according to any one of claims 1 to 5 when executing the computer program.

8. A computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program which, when executed by a processor, causes the processor to execute the information recommendation method based on the problem contribution degree according to any one of claims 1 to 5.