CN112989217B - System for managing human veins - Google Patents
System for managing human veins Download PDFInfo
- Publication number
- CN112989217B CN112989217B CN202110212854.6A CN202110212854A CN112989217B CN 112989217 B CN112989217 B CN 112989217B CN 202110212854 A CN202110212854 A CN 202110212854A CN 112989217 B CN112989217 B CN 112989217B
- Authority
- CN
- China
- Prior art keywords
- module
- data
- user
- preset
- relationship
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9536—Search customisation based on social or collaborative filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/01—Social networking
Abstract
The application provides a system for personal management. Aims to enable users to comprehensively master self-interpersonal resources, manage and maintain the prior interpersonal resources in the angles of circles and small groups, and expand the interpersonal resources. The system comprises a model training module, and specifically comprises: the data collection module is used for acquiring imprinting data between a user and a plurality of friends; the big data index calculation module is used for establishing a plurality of big data index systems according to the marked data; the first data set generating module is used for establishing a first data set by taking each big data index system as a sample, and dividing the first data set into a first training set and a first testing set according to a first preset proportion; and the prediction model training module is used for training various prediction models by using the first training set to obtain various initial prediction models, verifying the various initial prediction models by using the first test set, and screening the initial prediction model with the highest accuracy from the various initial prediction models as the relationship strength prediction model.
Description
Technical Field
The invention relates to the technical field of information, in particular to a system for managing human veins.
Background
Nowadays, with the development of the times, the relationships among people are more and more emphasized in daily life, work and study, the human channels become indispensable resources in the development process of individuals, and the effective management of the existing human channels of the individuals and the widening of the relationship among the human channels according to the existing human channels become increasingly important.
However, in the prior art, the management of the personal pulse mainly aims at the work field, a platform for daily personal pulse management is not provided, the relationship strength and the like are not divided more finely, and the personal pulse is not grasped comprehensively. Secondly, the prior art lacks of managing and maintaining the view angle of the existing interpersonal relationship in terms of circles and small groups, and the user cannot accurately grasp the information of the resources of the user, the different circles in which the user is located, the main topics of the circles or small groups, the positions of the user in the different circles, the overlapping degree between the circles and the like. Finally, the management software of the prior art, which is mostly point-to-point personal fission, lacks a larger fission for realizing the increase of the user opportunity from the perspective of circles and small groups. The user cannot explore some topics in weak connection or indirect connection, cannot know which people to find when wanting to enter a new circle or explore new resources, the positions of the middlemen in the circle or small group, which opinion leaders the circle has, and the like.
Disclosure of Invention
In view of this, the invention provides a system for managing human veins, which aims to solve the problems that a user lacks comprehensive understanding on the human veins, lacks management and maintenance on the existing human veins from the perspective of circles and small groups, and lacks realization of the human veins expansion of the user from the perspective of circles and small groups.
The embodiment of the application provides a system for people management, the system for people management includes a model training module, the model training module specifically includes:
the data collection module is used for acquiring imprinting data between the user and the friends and acquiring the actual relationship strength between the user and the friends;
the big data index calculation module is used for establishing a plurality of big data index systems according to the imprinting data between the user and the friends, wherein the big data index system consists of a plurality of big data indexes for representing the point-to-point relation strength between the user and a single friend;
the first data set generating module is used for establishing a first data set by taking each big data index system as a sample, dividing the first data set into a first training set and a first testing set according to a first preset proportion, and taking the actual relationship strength between friends corresponding to the sample and the user as a label of the sample;
and the prediction model training module is used for training a plurality of prediction models by using the first training set to obtain a plurality of initial prediction models, verifying the plurality of initial prediction models by using the first testing set, and screening out the initial prediction model with the highest accuracy from the plurality of initial prediction models as a relation strength prediction model, wherein the relation strength prediction model is used for predicting the relation strength between the user and each friend.
The embodiment of the application has the following advantages:
on the first hand, according to the social imprinting data of the user, the relationship strength between the user and each friend can be obtained through a relationship strength prediction model obtained through training, a user relationship graph with strong, acquaintance, weak and indirect relationships is established, and suggestions such as different priorities for setting messages to be sent and received, privacy level setting and the like are provided for the user according to different relationship strengths, so that the user can have more comprehensive grasp on the self relationship.
According to the second aspect, different circles can be established according to social imprinting data of the user through the system for managing the interpersonal relationship, different relationship networks corresponding to the circles are established according to the different relationship networks, and information such as the circles where the user is located, main topics of the circles, the position where the user is located in the circles, the overlapping degree of the circles where the user is located and the like is obtained through calculation, so that the user can manage and maintain the existing interpersonal relationship in the angles of the circles and small groups, and the existing resources can be better maintained and stabilized.
And thirdly, establishing different circles according to the social imprinting data of the user through an opinion leader identification model in the system for managing the human veins and the system for managing the human veins, establishing different relation networks corresponding to the circles according to the different circle identification model, calculating to obtain the positions of the members in the different relation networks in the respective circles, and obtaining the opinion leaders of the circles according to the opinion leader identification model, so that the human veins are expanded by the user from the angles of the circles and small groups, and the user enters the different circles through establishing a connection with other members through a middle person in each circle and establishing a connection with the opinion leaders of the circles.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the description of the embodiments of the present application will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art that other drawings can be obtained according to these drawings without inventive exercise.
FIG. 1 is a block diagram illustrating a system for people management according to an embodiment of the present application;
FIG. 2 is a flow diagram illustrating a predictive model training module according to an embodiment of the present application;
FIG. 3 is a flow diagram illustrating an opinion leader identification model training module according to an embodiment of the present application;
fig. 4 is a block diagram illustrating an overall architecture of a system for personal management according to an embodiment of the present application.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are some, but not all, embodiments of the present application. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
Before explaining the system for personal management provided in the present application, first, the personal management in the related art will be briefly explained below. The management of the human arteries in the prior art is mainly directed at the work field, a platform for daily individual human artery management is not provided, the relationship strength and the like are not divided more finely, the personal human arteries are lack of comprehensive grasp, secondly, the visual angle of the existing human arteries is lack of management and maintenance in the prior art in the angles of circles and small groups, and meanwhile, management software for human artery fission in the prior art is mostly point-to-point personal fission and lacks of larger fission for realizing the increase of the user opportunity in the angles of circles and small groups.
Therefore, the system for managing the personal veins is provided for overcoming the problems that the user lacks comprehensive grasp on the personal veins, the prior personal veins are managed and maintained in a circle and small group angle and the personal veins of the user are expanded in a circle and small group angle in the related technology, and the system for managing the personal veins is provided, so that the system not only can establish the user personal veins with strong, acquaintance, weak and indirect connection relations, but also provides suggestions for setting different priorities of sending and receiving messages, privacy level setting and the like for the user aiming at different relation strengths, and the user has more comprehensive grasp on the personal veins; the user can manage and maintain the existing human veins in the angles of circles and small groups, so that the existing resources can be maintained and stabilized better; the user expands the interpersonal relationship from the perspective of circles and small groups, and the user establishes contact with other members through the middlers in each circle to enter different circles and establishes contact with the opinion leaders of each circle to enter different circles.
Fig. 1 is a block diagram illustrating a system for personal management according to an embodiment of the present application. Referring to fig. 1, a system 100 for personal management provided by the present application includes a model training module 101, which specifically includes: the data collection module 102 is configured to obtain imprinting data between a user and a plurality of friends and obtain actual relationship strengths between the user and the plurality of friends; the big data index calculation module 103 is configured to establish a plurality of big data index systems according to the imprinted data between the user and the multiple friends, where the big data index system is composed of multiple big data indexes representing the strength of the point-to-point relationship between the user and a single friend; a first data set generating module 104, configured to use each big data index system as a sample, establish a first data set, and divide the first data set into a first training set and a first testing set according to a first preset proportion, where an actual relationship strength between a friend corresponding to the sample and the user is used as a label of the sample; the prediction model training module 105 is configured to train multiple prediction models with the first training set to obtain multiple initial prediction models, verify the multiple initial prediction models with the first testing set, and screen out an initial prediction model with the highest accuracy among the multiple initial prediction models as a relationship strength prediction model, where the relationship strength prediction model is to be used for predicting relationship strength between a user and each friend.
In this embodiment, the data collecting module 102 is configured to obtain imprinted data between the user and the multiple friends, and obtain actual relationship strength between the user and the multiple friends.
Illustratively, the imprinted data refers to chat data of messages sent by users in social software; the actual relationship strength refers to a real relationship between the user and each friend, for example, the user is in family relationship with the friend a, the user is in close acquaintance relationship with the friend B (e.g., girlfried, young, etc.), the user is in general acquaintance relationship with the friend C (e.g., occasionally contacted friends, general colleagues, etc.), the user is known to the user (e.g., mutual friends, people who have not sent interactive messages basically), and the actual relationship strength is determined by questionnaire survey, etc. The relationship strength specifically includes five layers including: family, pseudo family; close and acquainted people; the general acquaintance; a potential acquaintance; the cognizant.
In the data collection module 102, after the authorization of the user, the data provider encrypts and uploads the imprinting data of the user to the distributed database through the data uploading end, so as to obtain the imprinting data of the social software background user interaction. The marked data is provided with a time stamp, is preprocessed chatting data and does not relate to specific chatting content, and is also provided with a unique identification ID of a user in social software, the time of the user and each friend forming a friend, a GPS position information record with the time stamp and the like. In order to make the obtained imprinting data between the user and the friends to train the prediction model more in line with the practical situation, the obtained imprinting data between the user and the friends to be used for training the prediction model is represented by the innermost layer (namely family and family-like): the middle three layers (close acquaintance; general acquaintance; potential acquaintance): the recognized person is 3:5: the ratio of 8 is obtained because, as a relationship of the pulse, family members more closely are fewer than acquaintances, and acquaintances are more than acquaintances. And finally, matching the unique identification ID of the user and each of the plurality of friends with the determined actual relationship strength between the user and the plurality of friends, for example, taking one friend as an example, the user has one unique identification ID, the friend has one unique identification ID, and the unique identification ID of the user is matched with the unique identification ID of the friend and the actual relationship strength between the user and the friend. And cleaning the obtained imprinting data after matching and storing the data in the csv file.
In this embodiment, the big data index calculation module 103 is configured to establish a plurality of big data index systems according to the imprinted data between the user and the plurality of friends, where the big data index system is composed of a plurality of big data indexes representing the strength of the point-to-point relationship between the user and a single friend.
Illustratively, the big data index characterizes the strength of a point-to-point relationship between the user and the friend, and the big data index has a plurality of types, including: the method comprises the following steps of (1) associating frequency, reciprocity degree, intimacy degree, relationship persistence, similarity and structural indexes, wherein the relationship persistence is characterized by the time of becoming a friend; the reciprocal degree is represented by the frequency of sending red packets and gifting gifts; the contact frequency is characterized by the message frequency sent by the user and the friend and the interaction standard deviation; the intimacy degree is characterized by the meeting frequency and the intimacy degree embodied by the grouping remarks of the friend list. Specifically, the following table 1 shows.
The big data index system is a system composed of all the big data indexes in table 1. The relationship strength between the user and each friend can be obtained through the established point-to-point big data index system between the user and the single friend, namely, the point-to-point big data index system between the user and the single friend can be established according to the marking data of the user and the single friend, and the relationship strength between the user and the single friend can be evaluated through the big data index system. Where point-to-point refers to the point-to-point between the user's point and the point of a single buddy.
In the big data index calculation module 103, after the matched and cleaned imprinted data is calculated by the big data index calculation module 103, a plurality of point-to-point big data index systems between the user and a plurality of friends can be obtained, that is, one point-to-point big data index system can be corresponding between the user and one friend, and the user and the plurality of friends can respectively correspond to the plurality of point-to-point big data index systems.
TABLE 1
In this embodiment, the first data set generating module 104 is configured to use each big data index system as a sample, establish a first data set, and divide the first data set into a first training set and a first testing set according to a first preset proportion, where an actual relationship strength between a friend corresponding to the sample and the user is used as a label of the sample.
For example, a big data index system of point-to-point between a user and a single friend is used as a sample for training a prediction model, and a plurality of samples represented by a plurality of big data index systems of point-to-point between the user and a plurality of friends can establish a big data set, namely a first data set. Then, the first data set is divided into a first training set and a first testing set according to a certain proportion, namely a first preset proportion. The samples in the first data set are point-to-point big data index systems between the user and the single friend, and the labels of the samples are actual relationship strengths of the user and the corresponding single friend as labels.
In the first data set generating module 104, the plurality of big data index systems are used as samples, and after being processed by the first data set generating module, the plurality of big data indexes are divided into a first training set and a first testing set according to a first preset proportion.
In this embodiment, the prediction model training module 105 is configured to train multiple prediction models with the first training set to obtain multiple initial prediction models, verify the multiple initial prediction models with the first testing set, and screen out an initial prediction model with the highest accuracy among the multiple initial prediction models as a relationship strength prediction model, where the relationship strength prediction model is to be used for predicting relationship strength between a user and each friend.
FIG. 2 is a flow diagram of a predictive model training module according to an embodiment of the present application. Referring to fig. 2, the management node of the relationship strength prediction model identifies whether to generate a first data set, stores the first data set locally after the generation, and divides the first data set into a first training set and a first testing set according to a first preset proportion, preferably 8:2 as a first preset ratio. And the first training set is used for training various prediction models to obtain various initial prediction models which are trained, the various initial prediction models are detected by the first testing set, and the initial prediction model with the highest accuracy is determined to be used as the relation strength prediction model.
Illustratively, the plurality of predictive models includes: supervised machine learning, support vector machine model (SVM), decision Tree (Decision Tree), logistic regression, random Forest (Random Forest) algorithm, integrated algorithm Gradient Boosting Tree model (XGBoost), and the like.
In the prediction model training module 105, training a plurality of prediction models by using a first test set to obtain a plurality of initial relationship strength prediction models corresponding to the plurality of prediction models; and then, verifying the multiple initial relationship strength prediction models by using the first test set, determining an initial relationship strength prediction model with the highest prediction accuracy in the multiple initial relationship strength prediction models, and taking the initial relationship strength prediction model as a relationship strength prediction model for finally predicting the relationship strength between the user and the friend.
In the present embodiment, for convenience of understanding, it is only stated that the prediction model is trained based on the imprinted data between one user and a plurality of friends. In the present application, the prediction model is not trained based on the imprinting data between one user and a plurality of friends, but may be trained based on the imprinting data between a plurality of users and a plurality of friends, for example, the prediction model is trained based on the total imprinting data such as the imprinting data between user 1 and a plurality of friends, the imprinting data between user 2 and a plurality of friends, and the imprinting data between user 3 and a plurality of friends.
In the embodiments of the present application, the relationship strength has five layers, including: family members, quasi-family members; close and acquainted people; the general acquaintance; a potential acquaintance; the person who knows it. The relationship strength prediction model obtained by training is a five-layer prediction model. In practical applications, the relationship strength prediction model may be a four-layer or three-layer prediction model. Combining the five layers of relationship strengths into four or three layers, for example, combining family, family-like and close-acquaintance into one layer, and not combining the other three layers; and combining close acquaintances, general acquaintances and potential acquaintances into one layer, and not combining the other two layers, and the like, so as to train the prediction model to obtain a four-layer or three-layer relationship strength prediction model. For the relationship strength prediction models of five layers, four layers and three layers, the relationship strength prediction model of five layers is the most strict prediction model among the three, and the absolute accuracy of different classification modes cannot be directly compared. Therefore, samples are subjected to oversampling (Over Sampling) processing before the prediction Model is trained to solve the problem of imbalance of each type of samples, the accuracy rates of random guessing of the relationship strength prediction models of the five layers, the four layers and the three layers are respectively 20%, 25% and 33%, the random prediction Model is used as a reference Model (Benchmark Model), and the accuracy rate of the relationship strength prediction models of the five layers, the four layers and the three layers, which is improved compared with the reference Model, is used as an evaluation standard of an optimal Model.
Specifically, the XGboost prediction model is preferably trained to obtain the relationship strength prediction model. Preferably, 80% of the first data set is used as the first training set and 20% of the first data set is used as the first test set, so that over-fitting and under-fitting can be avoided, and the accuracy of the relationship strength prediction model is kept high. Preferably, in the training process, using an xgbclasifier method in an XGBoost library in Python language, taking the prediction accuracy of five-fold cross validation as an evaluation index, adjusting basic parameters of XGBoost, and finally training the XGBoost prediction model with a parameter combination of { learning rate (learning _ rate) per time of promotion being 0.1, depth (max _ depth) of tree being 3, number of trees (n _ estimators) being 550} and other parameters being default values, the finally obtained relationship strength prediction model being a four-layer prediction model, the four layers including: family/pseudo-family, close acquaintance, general acquaintance, potential acquaintance, and acquaintance.
In this application, the system for managing human veins further includes a prediction model modification module 106, which specifically includes:
the contribution degree determining module 1061 is configured to analyze, through an interpretable model, each big data index for training the relationship strength prediction model to obtain a contribution degree of each big data index to the relationship strength prediction model; the regression analysis module 1062 is configured to analyze, through a regression analysis model, each big data index for training the relationship strength prediction model to obtain a correlation relationship of each big data index in the relationship strength prediction model; the wrong prediction path determining module 1063 is configured to analyze, through a decision path diagram, a wrong prediction result in a process of verifying an accuracy rate of the relationship strength prediction model, to obtain a wrong prediction path of the wrong prediction result; an updating module 1064, configured to update the multiple big data index systems according to the contribution degree, the correlation and the misprediction path.
Illustratively, the interpretable model is SHAP, which is a model-interpreted package developed by Python, which can interpret the output of the model. In a contribution degree determining module, the contribution degree of each big data index in a big data index system to the accuracy rate of the relation strength prediction model is obtained by importing the obtained relation strength prediction model into an SHAP library; in the regression analysis module, the obtained relationship strength prediction model is analyzed through the regression analysis model, and the correlation relationship, namely the relevance, among all big data indexes in the big data index system can be obtained; and finally, in an error prediction path determining module, in the process of verifying the accuracy of the obtained relation strength prediction model, an error prediction result exists, and the error prediction result is analyzed through a decision path diagram to obtain an error prediction path. And when the contribution degree of each big data index in the obtained big data index system to the accuracy rate of the relation strength prediction model is lower than a certain preset value, removing the contribution degree from the established point-to-point big data index system between the user and the single friend. In an updating module, according to the contribution degree of each big data index to the accuracy rate of the relation strength prediction model and whether each big data index has relevance and an incorrect prediction path, a big data index system is updated, whether the big data index in the original big data index system needs to be deleted or other big data indexes need to be added to the original big data index is determined, and therefore the big data index system is updated.
Specifically, the maximum contribution of the link frequency in the big data index system to the overall prediction accuracy of the relation strength model is determined according to the contribution degree of each big data index in the big data index system determined by the contribution degree determination module to the accuracy of the relation strength prediction model, the correlation among the big data indexes obtained by the regression analysis module, and the wrong prediction path obtained by the wrong prediction path determination module. And partial big data indexes do not contribute greatly to the overall accuracy of the relationship strength prediction model, but have important significance in identifying a specific relationship strength circle layer. For example, the contribution degree determining module, the regression analysis module and the error prediction path determining module analyze that the message frequency in the non-working time contributes most to the prediction accuracy of the second-layer acquaintances of the relationship strength prediction model; the gender similarity index is favorable for identifying a second circle layer, and the second circle layer is mostly of the tiegons and the girlfriends, namely people prefer to establish close and acquaintance relations with people with the same gender; the age similarity and the industry similarity are beneficial to identifying the third circle, and the third circle is mostly related to classmates and colleagues, so the age and the industry have important significance for identifying the circle. And after the big data index system is updated, the relation strength prediction model is trained again by the updated big data index system to obtain an optimized relation strength prediction model, namely the first relation strength prediction model. By using the interpretable model for analysis, after the big data index of the meeting frequency of dividing the holiday and the non-holiday from the time dimension is obtained, the accuracy of the first relation strength prediction model is improved compared with the accuracy of the relation strength prediction model. 8. The importance of the frequent big data indexes of September is higher than that of the general frequent big data indexes of September, and the importance of the frequent big data indexes of national celebration and holiday is relatively low, wherein the accuracy contribution of August to the relation strength prediction model for predicting the family of the innermost layer is the largest, and the accuracy contribution of September to the relation strength prediction model for predicting the second-layer acquaintance and the person known to the outermost layer is the largest.
The data set updating module 1065 is configured to establish a modified data set by using the updated large data index systems as samples, and divide the modified data set into a modified training set and a modified testing set according to the first preset proportion, where the actual relationship strength between the friend corresponding to the sample and the user is used as a label of the sample; and the correcting module 1066 is configured to train the relationship strength prediction model with the correction training set to obtain a second relationship strength prediction model, test the second relationship strength prediction model with the correction test set, and when the accuracy of the second relationship strength prediction model meets a first preset threshold, use the second relationship strength prediction model as a first relationship strength prediction model, where the first relationship strength prediction model is a prediction model obtained by optimizing the relationship strength prediction model.
Illustratively, in the data set updating module, after the big data index system is updated, a part of big data indexes in the big data index system are deleted, or new big data indexes are added, each updated big data index is used as a sample, a modified data set is established, and then the modified data set is divided into a modified training set and a modified testing set according to a first preset proportion. The label of the sample is also the actual relationship strength between the user and the corresponding single friend as the label. In the correction module, the correction training set is used for training the obtained relationship strength prediction model again to obtain a second relationship strength prediction model, the accuracy of the second relationship strength prediction model is verified, and when the accuracy is higher than a set first preset threshold value, the second relationship strength prediction model is used as a first relationship strength prediction model which is an optimized relationship strength prediction model and is used for predicting the relationship strength between the user and each friend.
In this application, the system for personal management further comprises: the updating module is used for updating the big data index systems and the multiple prediction models according to the contribution degree, the correlation and the error prediction path; and the correction module is used for training the updated multiple prediction models by using the correction training set to obtain multiple correction prediction models, verifying the multiple correction prediction models by using the correction test set, and screening out the correction prediction model with the highest accuracy from the multiple correction prediction models as a first relation strength prediction model.
For example, in the present application, the updating module may update not only the big data index system, but also multiple prediction models, that is, after the big data index system is updated, training the previous multiple prediction models based on the updated big data index system or training the previous multiple prediction models and additionally added prediction models based on the updated big data index system. For the specific implementation of updating the big data index system in this embodiment, refer to the specific implementation of the contribution degree determining module, the regression analysis module, and the error prediction path determining module, which is not described herein again; for a specific implementation of the update of the data set, reference is made to the above specific implementation of the data set update module, which is not described herein again. In the correction module, the correction training set is used for training the updated multiple prediction models to obtain multiple correction prediction models, then the accuracy of the multiple prediction models is verified through the correction testing set, and the model with the highest accuracy is selected as the first relation strength prediction model.
In this application, the system for personal management further comprises: the data collection module 102 is configured to obtain imprinted data of a group where a user is located; a fixed theme generating module 107, configured to generate a plurality of fixed themes according to the imprinting data between the user and the plurality of friends and the imprinting data of the group in which the user is located; a circle dividing module 108, configured to divide, according to the fixed topics, a plurality of circles that have a one-to-one correspondence relationship with the fixed topics; a relationship network establishing module 109, configured to establish a relationship network of each of the circles according to the imprinted data between the user and the multiple friends and the imprinted data of the group in which the user is located; a network index calculation module 110, configured to establish a respective network index system of each member in the relationship network according to the respective relationship network of the plurality of circles, where the respective network index system of each member in the relationship network includes a plurality of network indexes; a second data set generating module 111, configured to use the respective network index systems of the members in the relationship network as samples, establish a second data set, and divide the second data set into a second training set and a second testing set according to a second preset proportion, where whether a member corresponding to a sample is an opinion leader is used as a label of the sample; and an opinion leader identification model training module 112, configured to train a model with the second training set to obtain an initial opinion leader identification model, verify the initial opinion leader identification model with the second test set, and when the accuracy of the initial opinion leader identification model meets a second preset threshold, use the initial opinion leader identification model as an opinion leader identification model, where the opinion leader identification model is used to identify opinion leaders in a relationship network.
In this embodiment, the data collection module also collects the imprinted data of the group where the user is located. The fixed theme generation module analyzes and processes the imprinting data between the user and the friends and the imprinting data of the group where the user is located to obtain a plurality of fixed themes, such as working themes, student themes, friends themes, family themes and the like. And the circle dividing module establishes a plurality of circles corresponding to the plurality of fixed topics according to the imprinting data between the user and the plurality of friends and the imprinting data of the group where the user is located on the basis of the plurality of fixed topics, wherein members in the plurality of circles comprise the user, the friends of the user and the group members of the group where the user is located. And the relationship network establishing module is used for establishing a relationship network corresponding to the circles through the imprinting data between the user and the friends and the imprinting data of the group where the user is located based on the obtained circles, wherein the relationship network is a network established according to the incidence relation among the members in one circle.
The network indexes include: degree centrality, intermediate centrality betweenness, proximity centrality cycloseness, pageRank, structural hole. The network index system is a system composed of the plurality of different network indexes and comprises the plurality of different network indexes. A member of a relationship network has its own network index hierarchy. And the network index calculation module is used for calculating according to the established multiple relationship networks and solving to obtain the respective network index systems of the members in each relationship network.
In the second data set generation module, the relationship network system of each member is used as a sample for training the recognition model, the relationship networks and the relationship network systems of the members in each relationship network form a second data set, the second data set is divided into a second training set and a second testing set according to a second preset proportion, for the network index system used as the sample, whether the member in the relationship network corresponding to the network index system is an opinion leader or not is used as a label, and the opinion leader is a member playing a leading role in a circle and is at least one. In the opinion leader recognition model training module, training a recognition model by using a second training set to obtain an initial opinion leader recognition model, verifying the accuracy of the initial opinion leader recognition model by using a second test set, taking the initial opinion leader recognition model as an opinion leader recognition model when the initial opinion leader recognition model meets a set second preset threshold, increasing the richness of the second training set when the initial opinion leader recognition model does not meet the set second preset threshold, obtaining imprinting data of more users, obtaining more big data index systems based on the obtained imprinting data of more users, adding the obtained big data index systems into the second training set, updating the second training set, training the recognition model by using the updated second training set until the accuracy of the obtained initial opinion leader recognition model meets the second preset threshold. The opinion leader identification model will be used to identify opinion leaders in a relational network.
In this application, the fixed topic generation module 107 includes: the content preprocessing module 1071 is configured to preprocess the imprinted data between the user and the multiple friends and the imprinted data of the group where the user is located, and includes: performing Chinese word segmentation on the imprinted data between the user and a plurality of friends and the imprinted data of the group where the user is located to obtain first data; removing stop words from the first data to obtain second data; a theme generation module 1072, configured to generate a plurality of fixed themes according to the second data.
Illustratively, in the content preprocessing module, after performing chinese word segmentation on the imprinted data between the user and the multiple friends and the imprinted data of the group in which the user is located, the chinese word segmentation is a process of recombining a sequence of consecutive words into a word sequence according to a certain specification. The stop words are removed from the marked data of the Chinese participles and used as second data, the stop words can be understood as some irrelevant words in the sentence, and the words can be filtered before the data is processed. And in the theme generation module, analyzing the second data through the LDA theme model to obtain a plurality of fixed themes.
Fig. 3 is a flowchart illustrating an opinion leader identification model training module according to an embodiment of the present application. Referring to fig. 3, an opinion leader identification model management node monitors whether a fixed theme is generated or not, when a plurality of fixed themes are generated, circles are divided according to different fixed themes to obtain a plurality of circles corresponding to the themes, and a relationship network is established with the topic circles as a principle, wherein one circle corresponds to one relationship network, and the relationship network comprises a plurality of members. And calculating the network indexes of the members in each relationship network, and establishing a data set according to the network indexes of the members, wherein the data set comprises a second training set and a second testing set which are divided according to a second preset proportion. And training the model by using the second training set to obtain an initial opinion leader recognition model, verifying the accuracy of the initial opinion leader recognition model by using the second test set, and when the accuracy is higher than a second preset threshold value, using the initial opinion leader recognition model as an opinion leader recognition model for recognizing the opinion leaders in the relational network.
In this application, fig. 4 is a block diagram illustrating an overall architecture of a system for personal management according to an embodiment of the present application. Referring to fig. 4, the system for managing personal connections further includes a user application module 201, which specifically includes: the data collection module 102 is configured to obtain imprinting data between a current user and each friend; a big data index calculation module 103, configured to calculate a big data index in each big data index system between the current user and each friend according to the imprinted data between the current user and each friend; a relationship strength prediction module 202, configured to input the big data indexes in each big data index system between the current user and each friend into the first relationship strength prediction model, so as to obtain the predicted relationship strength between the current user and each friend; the relationship map module 203 is configured to establish a relationship map of the current user according to the strength of the predicted relationship between the current user and each friend, and visualize the relationship map of the current user according to the relationship among strong links, acquaintances, weak links, and indirect links.
Illustratively, a current user requests a face atlas establishing function, a face atlas establishing function management node is triggered, a data collecting module is started, imprinted data between the current user and each friend is obtained, the obtained imprinted data is matched and cleaned and then stored in a csv file, the specific matching implementation mode is the same as that of the imprinted data of the data collecting module in the model training module, and details are not repeated here. And the big data index calculation module is used for calculating the marked data between the matched and cleaned current user and each friend to obtain big data indexes in each big data index system between the current user and each friend, wherein the big data index types included in the big data index system between the user and a single friend are the big data index types included in the updated big data index system for training the first relation strength prediction model. And the relationship strength prediction module is used for inputting the big data indexes in each big data index system between the current user and each friend into the trained first relationship strength prediction model to obtain the prediction relationship strength between the current user and each friend. And the personal map module is used for establishing the personal map of the current user according to the prediction relationship strength between the current user and each friend. The relationship between each friend and the current user in the pulse-beat map is the relationship between the strong connection (family/family-like), acquaintance (close acquaintance), weak connection (general acquaintance) and indirect connection (potential acquaintance/known acquaintance), and the pulse-beat map is displayed to the current user in a visualized manner. The current user can set different privacy authorities for friends in different circles and set different message processing priorities for friends in different circles according to the popularity map, and the popularity management system can provide suggestions for friend transfer introduction.
In this application, the user application module further includes: the data collection module 102 is configured to obtain imprinted data of a group where a current user is located; a fixed theme generating module 107, configured to generate a plurality of preset fixed themes according to the imprinting data between the current user and each friend and the imprinting data of the group where the current user is located; a circle dividing module 108, configured to divide, according to the multiple preset fixed topics, multiple preset circles that have a one-to-one correspondence relationship with the multiple preset fixed topics; a relationship network establishing module 109, configured to establish a relationship network of each of the preset circles according to the imprinting data between the current user and each friend and the imprinting data of the group where the current user is located; a network index calculating module 110, configured to calculate a network index of the current user in the multiple preset circles according to respective relationship networks of the multiple preset circles; the human-pulse management module 204 is configured to obtain, according to the network index of the current user in the preset circles, a position of the current user in the relationship network of each of the preset circles and an overlap degree between the preset circles, output a division result of the preset circles, a network feature of each of the preset circles, and visually display the position and the overlap degree.
Exemplarily, the current user requests a personal resource management function, triggers a personal resource management function management node, starts a data collection module, and obtains marking data of a group where the current user is located and marking data between the current user and each friend. And the fixed theme generating module is used for generating a plurality of preset fixed themes according to the imprinting data between the current user and each friend and the imprinting data of the group where the current user is located. The circle dividing module divides a plurality of preset circles corresponding to the preset fixed themes according to the imprinting data between the current user and each friend and the imprinting data of the group where the current user is located on the basis of the preset fixed themes, wherein one preset fixed theme corresponds to one preset circle, and the preset fixed theme and the preset circle are in one-to-one correspondence. And the relationship network establishing module is used for establishing respective relationship networks of the preset circles according to the marking data between the current user and each friend and the marking data of the group where the current user is located on the basis of the preset circles. And the network index calculation module is used for calculating the network indexes of the current user in the network index systems in the preset circles according to the relationship networks of the preset circles, so that the specific position of the current user in the relationship networks in the preset circles can be determined. The personal management module obtains the position of the current user in the relationship network of each preset circle and the overlapping degree between the preset circles according to the network indexes of the current user in each preset circle, wherein the overlapping degree comprises two conditions, one is that the members are positioned in the preset circles at the same time, the preset circles are considered to be overlapped with each other, when the number of the members positioned in the preset circles at the same time is more, the overlapping degree of the preset circles is considered to be higher, and the other is that the members positioned between the different preset circles have friend relationships or have contact. Outputting the division results of the preset circles and the respective network characteristics of the preset circles, wherein the network characteristics comprise: the network density, the network scale and the network clustering coefficient of the network formed by the preset circles visually display the specific position of the current user in the preset circles and the overlapping degree of the preset circles.
In this application, the user application module further includes: a network index calculating module 110, configured to calculate network indexes in the network index systems of the members in the preset circles according to the respective relationship networks of the preset circles; an opinion leader identification module 205, configured to input network indexes in the network index system of each member in the plurality of preset circles into the opinion leader identification model, so as to obtain opinion leaders of each of the plurality of preset circles; the human vein resource expansion module 206 is configured to obtain a plurality of preset circle maps corresponding to the preset circles according to the respective relationship networks of the preset circles and the respective opinion leaders of the preset circles, visually display the preset circle maps, and mark key members in the preset circle maps, where the key members are members that the current user enters the preset circle where the key members are located and that the opinion leaders of the preset circle where the key members are located need to be contacted.
Exemplarily, the current user requests a human resource expansion function, triggers the function management node, starts a data collection module, and obtains marking data of a group where the current user is located and marking data between the current user and each friend. And the fixed theme generating module is used for generating a plurality of preset fixed themes according to the imprinting data between the current user and each friend and the imprinting data of the group where the current user is located. The circle dividing module divides a plurality of preset circles corresponding to the preset fixed themes according to the imprinting data between the current user and each friend and the imprinting data of the group where the current user is located on the basis of the preset fixed themes, wherein one preset fixed theme corresponds to one preset circle, and the preset fixed theme and the preset circle are in one-to-one correspondence. And the relationship network establishing module is used for establishing respective relationship networks of the preset circles according to the marking data between the current user and each friend and the marking data of the group where the current user is located on the basis of the preset circles. And the network index calculation module is used for calculating network indexes in each member network index system in the relationship network of each preset circle according to the relationship network of each preset circle. And the opinion leader identification module is used for inputting the calculated network indexes in the member network index systems in the relationship network of each preset circle into the trained opinion leader identification model to determine the opinion leaders of each preset circle, wherein a plurality of opinion leaders can exist in one circle at the same time. The personal vein resource expansion module obtains a plurality of preset circle maps corresponding to a plurality of preset circles according to established relationship networks of the preset circles and opinion leaders of the preset circles, visually displays the preset circle maps, marks key members of the preset circles, wherein the key members can be members closest to the opinion leaders in the relationship networks of the preset circles, or members with the largest number of connections with other members in the relationship networks of the preset circles (for example, a relationship network has 5 members A, B, C, D, E, a has connections with the remaining four members, B has connections with the remaining four members, C has connections with A, B, D has connections with A, B, E, E has connections with A, B, D, and the number of the members with the largest number of connections with other members in the relationship network is 3536 zxft, and the number of the contacts with the key members in the relationship network is 3536, so that the number of the key members in the relationship network is equal to the number of the key members in the preset circle 3926. The current user can be quickly integrated into the preset circle where the key member is located through the key member, and the current user can establish contact with the opinion leader in the preset circle where the key member is located through the key member. It can be understood that, when the current user needs to enter a circle and an unknown pulse enters the circle, that is, when no known friend is in the circle and a member in the circle exists in the group of the current user, the current user can perform analysis by using the pulse management system in the application to find out the member in the circle in the group of the current user, and the member slowly merges into the circle; meanwhile, when a user needs to establish contact with the opinion leader of the circle after entering the circle and the current user cannot directly establish contact with the opinion leader of the circle, the current user can analyze through the system for managing the personal veins in the application to determine the member nearest to the opinion leader of the circle in the circle and establish contact with the opinion leader of the circle through the member.
The system for managing the human arteries can help users to effectively expand and explore more human resources. Besides maintaining the strong linkage relationship, the user can also expand the utilization of the weak linkage and indirect linkage resources of the group in which the user is located. The method comprises the steps of obtaining weak link chat topics and indirect link chat topics by using group chat content analysis, dividing different circles, finding an opinion leader and a circle with the opinion leader as a core according to an interaction rule, effectively knowing topics interested by a certain circle when a user wants to explore new resources or enter the new circle, finding corresponding individuals as bridges to help the individuals to enter the circle, knowing how the individuals can be connected with the opinion leader of the circle, and the like, and realizing efficient series connection of resources.
The embodiments in the present specification are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other.
Finally, it should also be noted that, in this document, relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or terminal that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or terminal. Without further limitation, an element defined by the phrase "comprising a … …" does not exclude the presence of another identical element in a process, method, article, or terminal apparatus that comprises the element.
The system for managing human arteries provided by the invention is described in detail above, and the principle and the implementation mode of the invention are explained by applying specific examples in the text, and the description of the examples is only used for helping to understand the method and the core idea of the invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present invention.
Claims (8)
1. The system for managing the human pulse is characterized by comprising a model training module, wherein the model training module specifically comprises:
the data collection module is used for acquiring imprinting data between the user and the friends and acquiring the actual relationship strength between the user and the friends;
the big data index calculation module is used for establishing a plurality of big data index systems according to the marked data between the user and the friends, wherein the big data index systems are composed of a plurality of big data indexes for representing the point-to-point relation strength between the user and a single friend;
the first data set generating module is used for establishing a first data set by taking each big data index system as a sample, and dividing the first data set into a first training set and a first testing set according to a first preset proportion, wherein the actual relationship strength between friends corresponding to the sample and the user is taken as a label of the sample;
and the prediction model training module is used for training a plurality of prediction models by using the first training set to obtain a plurality of initial prediction models, verifying the plurality of initial prediction models by using the first testing set, and screening out the initial prediction model with the highest accuracy from the plurality of initial prediction models as a relation strength prediction model, wherein the relation strength prediction model is used for predicting the relation strength between the user and each friend.
2. The system of claim 1, wherein the model training module further comprises: a prediction model modification module, the prediction model modification module specifically comprising:
the contribution degree determining module is used for analyzing each big data index for training the relation strength prediction model through an interpretable model to obtain the contribution degree of each big data index to the relation strength prediction model;
the regression analysis module is used for analyzing each big data index for training the relationship strength prediction model through a regression analysis model to obtain the correlation relationship of each big data index in the relationship strength prediction model;
the error prediction path determining module is used for analyzing an error prediction result in the process of verifying the accuracy of the relationship strength prediction model through a decision path diagram to obtain an error prediction path of the error prediction result;
the updating module is used for updating the big data index systems according to the contribution degree, the correlation and the error prediction path;
the data set updating module is used for establishing a correction data set by taking each updated big data index system as a sample, and dividing the correction data set into a correction training set and a correction testing set according to the first preset proportion, wherein the actual relationship strength between friends corresponding to the sample and the user is taken as a label of the sample;
and the correction module is used for training the relation strength prediction model by using the correction training set to obtain a second relation strength prediction model, testing the second relation strength prediction model by using the correction testing set, and when the accuracy of the second relation strength prediction model meets a first preset threshold, taking the second relation strength prediction model as a first relation strength prediction model which is a prediction model obtained after the relation strength prediction model is optimized.
3. The system of claim 2, wherein the predictive model modification module further comprises:
the updating module is used for updating the big data index systems and the multiple prediction models according to the contribution degree, the correlation and the error prediction path;
and the correction module is used for training the updated multiple prediction models by using the correction training set to obtain multiple correction prediction models, verifying the multiple correction prediction models by using the correction test set, and screening out the correction prediction model with the highest accuracy from the multiple correction prediction models as a first relation strength prediction model.
4. The system of claim 1, further comprising:
the data collection module is used for acquiring imprinting data of a group where a user is located;
the fixed theme generating module is used for generating a plurality of fixed themes according to the imprinting data between the user and the friends and the imprinting data of the group where the user is located;
the circle dividing module is used for dividing a plurality of circles which have one-to-one correspondence with the plurality of fixed themes according to the plurality of fixed themes;
the relation network establishing module is used for establishing respective relation networks of a plurality of circles according to the marking data between the user and the friends and the marking data of the group where the user is located;
the network index calculation module is used for establishing respective network index systems of the members in the relationship network according to the respective relationship networks of the circles, wherein the respective network index system of each member in the relationship network comprises a plurality of network indexes;
a second data set generating module, configured to use the respective network index systems of the members in the relationship network as samples, establish a second data set, and divide the second data set into a second training set and a second testing set according to a second preset proportion, where whether a member corresponding to a sample is an opinion leader is used as a label of the sample;
and the opinion leader identification model training module is used for training a model by using the second training set to obtain an initial opinion leader identification model, verifying the initial opinion leader identification model by using the second testing set, and when the accuracy of the initial opinion leader identification model meets a second preset threshold, using the initial opinion leader identification model as an opinion leader identification model which is used for identifying opinion leaders in a relational network.
5. The system of claim 4, wherein the fixed topic generation module comprises:
the content preprocessing module is used for preprocessing the marking data between the user and the friends and the marking data of the group where the user is located, and comprises the following steps: performing Chinese word segmentation on the imprinted data between the user and the friends and the imprinted data of the group where the user is located to obtain first data; removing stop words from the first data to obtain second data;
and the theme generating module is used for generating a plurality of fixed themes according to the second data.
6. The system according to claim 3, wherein the system for people management further comprises a user application module, the user application module specifically comprising:
the data collection module is used for acquiring imprinting data between the current user and each friend;
the big data index calculation module is used for calculating big data indexes in each big data index system between the current user and each friend according to the marked data between the current user and each friend;
the relationship strength prediction module is used for inputting big data indexes in each big data index system between the current user and each friend into the first relationship strength prediction model to obtain the prediction relationship strength between the current user and each friend;
and the personal map module is used for establishing the personal map of the current user according to the strength of the predicted relationship between the current user and each friend and visualizing the personal map of the current user by using the relationship among strong links, acquaintances, weak links and indirect links.
7. The system of claim 5, wherein the user application module further comprises:
the data collection module is used for acquiring imprinting data of a group where the current user is located;
the fixed theme generating module is used for generating a plurality of preset fixed themes according to the imprinting data between the current user and each friend and the imprinting data of the group where the current user is located;
the circle dividing module is used for dividing a plurality of preset circles which have one-to-one correspondence with the preset fixed themes according to the preset fixed themes;
the relation network establishing module is used for establishing the relation networks of the preset circles according to the marking data between the current user and the friends and the marking data of the group where the current user is located;
the network index calculation module is used for calculating the network indexes of the current user in the preset circles according to the respective relationship networks of the preset circles;
and the human pulse management module is used for acquiring the position of the current user in the relationship network of each preset circle and the overlapping degree between the preset circles according to the network indexes of the current user in the preset circles, outputting the division results of the preset circles, the network characteristics of each preset circle, and visually displaying the position and the overlapping degree.
8. The system of claim 7, wherein the user application module further comprises:
the network index calculation module is used for calculating network indexes in a network index system of each member in the preset circles according to the respective relationship networks of the preset circles;
the opinion leader identification module is used for inputting network indexes in a network index system of each member in the preset circles into the opinion leader identification model to obtain opinion leaders of the preset circles;
the human vein resource expansion module is used for obtaining a plurality of preset circle maps corresponding to the preset circles according to the respective relationship networks of the preset circles and the respective opinion leaders of the preset circles, visually displaying the preset circle maps, and marking key members in the preset circle maps, wherein the key members are members which the current user enters the preset circle where the key members are located and which need to be contacted when the opinion leaders of the preset circle where the key members are located are recognized.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110212854.6A CN112989217B (en) | 2021-02-25 | 2021-02-25 | System for managing human veins |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110212854.6A CN112989217B (en) | 2021-02-25 | 2021-02-25 | System for managing human veins |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112989217A CN112989217A (en) | 2021-06-18 |
| CN112989217B true CN112989217B (en) | 2022-12-13 |
Family
ID=76350750
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110212854.6A Active CN112989217B (en) | 2021-02-25 | 2021-02-25 | System for managing human veins |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112989217B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116501979A (en) * | 2023-06-30 | 2023-07-28 | 北京水滴科技集团有限公司 | Information recommendation method, information recommendation device, computer equipment and computer readable storage medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109919684A (en) * | 2019-03-18 | 2019-06-21 | 上海盛付通电子支付服务有限公司 | For generating method, electronic equipment and the computer readable storage medium of information prediction model |
| CN111311321A (en) * | 2020-02-14 | 2020-06-19 | 北京百度网讯科技有限公司 | User consumption behavior prediction model training method, device, equipment and storage medium |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10796095B2 (en) * | 2018-04-05 | 2020-10-06 | Adobe Inc. | Prediction of tone of interpersonal text communications |
-
2021
- 2021-02-25 CN CN202110212854.6A patent/CN112989217B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109919684A (en) * | 2019-03-18 | 2019-06-21 | 上海盛付通电子支付服务有限公司 | For generating method, electronic equipment and the computer readable storage medium of information prediction model |
| CN111311321A (en) * | 2020-02-14 | 2020-06-19 | 北京百度网讯科技有限公司 | User consumption behavior prediction model training method, device, equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112989217A (en) | 2021-06-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Rodriguez et al. | A computational social science perspective on qualitative data exploration: Using topic models for the descriptive analysis of social media data | |
| Purohit et al. | Emergency-relief coordination on social media: Automatically matching resource requests and offers | |
| US10178197B2 (en) | Metadata prediction of objects in a social networking system using crowd sourcing | |
| CN117219080A (en) | Virtual assistant for generating personalized responses within a communication session | |
| US8949237B2 (en) | Detecting overlapping clusters | |
| US20120313948A1 (en) | Method and System for Providing Community Networking Services in an Organization | |
| CN109472305A (en) | Answer quality determines model training method, answer quality determination method and device | |
| JP6488753B2 (en) | Information processing method | |
| CN104731958A (en) | User-demand-oriented cloud manufacturing service recommendation method | |
| Maaloui et al. | “Cruel intention” or “entrepreneurial intention”: what did you expect? An overview of research on entrepreneurial intention—an interactive perspective | |
| Kalampokis et al. | On predicting election results using twitter and linked open data: the case of the UK 2010 election | |
| CN112989217B (en) | System for managing human veins | |
| Cheng | Data-mining research in education | |
| Sheeba et al. | A fuzzy logic based on sentiment classification | |
| CN110472057B (en) | Topic label generation method and device | |
| Marbouti et al. | Social Media Analyst Responding Tool: A Visual Analytics Prototype to Identify Relevant Tweets in Emergency Events. | |
| US11134045B2 (en) | Message sorting system, message sorting method, and program | |
| Imuede et al. | Sentiment analysis as an indicator to evaluate gender disparity on sexual violence tweets in South Africa | |
| Adel et al. | Towards socially intelligent automated tutors: Predicting learning style dimensions from conversational dialogue | |
| CN113313470A (en) | Employment type evaluation method and system based on big data | |
| Xie et al. | Multi-faceted classification for the identification of informative communications during crises: Case of Covid-19 | |
| Thiriot | Generate descriptive social networks for large populations from available observations: A novel methodology and a generator | |
| Becatti et al. | Collaboration and followership: a stochastic model for activities in bipartite social networks | |
| US20230214679A1 (en) | Extracting and classifying entities from digital content items | |
| Beaver et al. | On the care and feeding of virtual assistants: Automating conversation review with AI |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |