CN112685514A - AI intelligent customer value management platform - Google Patents
AI intelligent customer value management platform Download PDFInfo
- Publication number
- CN112685514A CN112685514A CN202110021661.2A CN202110021661A CN112685514A CN 112685514 A CN112685514 A CN 112685514A CN 202110021661 A CN202110021661 A CN 202110021661A CN 112685514 A CN112685514 A CN 112685514A
- Authority
- CN
- China
- Prior art keywords
- data
- platform
- analysis
- database
- class
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 claims abstract description 46
- 238000007726 management method Methods 0.000 claims abstract description 29
- 238000004458 analytical method Methods 0.000 claims abstract description 22
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 16
- 238000005516 engineering process Methods 0.000 claims abstract description 16
- 238000007405 data analysis Methods 0.000 claims abstract description 12
- 230000000007 visual effect Effects 0.000 claims abstract description 12
- 230000008676 import Effects 0.000 claims abstract description 4
- 238000010586 diagram Methods 0.000 claims description 26
- 238000012549 training Methods 0.000 claims description 20
- 230000008569 process Effects 0.000 claims description 16
- 238000013523 data management Methods 0.000 claims description 11
- 238000012360 testing method Methods 0.000 claims description 11
- 238000007418 data mining Methods 0.000 claims description 10
- 230000002452 interceptive effect Effects 0.000 claims description 8
- 230000006870 function Effects 0.000 claims description 7
- 238000012545 processing Methods 0.000 claims description 7
- 230000008901 benefit Effects 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000013461 design Methods 0.000 claims description 6
- 238000011161 development Methods 0.000 claims description 6
- 230000003993 interaction Effects 0.000 claims description 6
- 210000001503 joint Anatomy 0.000 claims description 5
- 238000007619 statistical method Methods 0.000 claims description 5
- 238000007621 cluster analysis Methods 0.000 claims description 4
- 239000000284 extract Substances 0.000 claims description 4
- 238000010801 machine learning Methods 0.000 claims description 4
- 238000005065 mining Methods 0.000 claims description 4
- 238000012800 visualization Methods 0.000 claims description 4
- 238000009825 accumulation Methods 0.000 claims description 3
- 230000000694 effects Effects 0.000 claims description 3
- 238000007689 inspection Methods 0.000 claims description 3
- 230000010354 integration Effects 0.000 claims description 3
- 230000007246 mechanism Effects 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 3
- 230000008520 organization Effects 0.000 claims description 3
- 230000002093 peripheral effect Effects 0.000 claims description 3
- 230000004044 response Effects 0.000 claims description 3
- 238000000926 separation method Methods 0.000 claims description 3
- 238000012300 Sequence Analysis Methods 0.000 claims description 2
- 238000007635 classification algorithm Methods 0.000 claims description 2
- 238000013145 classification model Methods 0.000 claims description 2
- 238000010276 construction Methods 0.000 claims description 2
- 238000010219 correlation analysis Methods 0.000 claims description 2
- 238000013500 data storage Methods 0.000 claims description 2
- 238000003066 decision tree Methods 0.000 claims description 2
- 230000003203 everyday effect Effects 0.000 claims description 2
- 239000006260 foam Substances 0.000 claims description 2
- 238000013507 mapping Methods 0.000 claims description 2
- 238000000611 regression analysis Methods 0.000 claims description 2
- 239000013589 supplement Substances 0.000 claims description 2
- 230000004083 survival effect Effects 0.000 claims description 2
- 238000012795 verification Methods 0.000 claims description 2
- 238000007794 visualization technique Methods 0.000 claims description 2
- 238000004891 communication Methods 0.000 description 4
- 230000002457 bidirectional effect Effects 0.000 description 3
- 230000015572 biosynthetic process Effects 0.000 description 3
- 230000001965 increasing effect Effects 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 238000003786 synthesis reaction Methods 0.000 description 3
- 238000012546 transfer Methods 0.000 description 3
- 238000006243 chemical reaction Methods 0.000 description 2
- 239000003795 chemical substances by application Substances 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 238000013136 deep learning model Methods 0.000 description 2
- 238000003032 molecular docking Methods 0.000 description 2
- 238000003058 natural language processing Methods 0.000 description 2
- 239000000047 product Substances 0.000 description 2
- 238000012384 transportation and delivery Methods 0.000 description 2
- 244000035744 Hura crepitans Species 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 230000004888 barrier function Effects 0.000 description 1
- 230000002354 daily effect Effects 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000035515 penetration Effects 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 230000002787 reinforcement Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 230000026676 system process Effects 0.000 description 1
- 238000013518 transcription Methods 0.000 description 1
- 230000035897 transcription Effects 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
Images
Abstract
The invention relates to the technical field of calling systems, in particular to an AI intelligent customer value management platform which integrates three main capabilities of a multi-channel entrance, an AI robot outbound and big data analysis technology; the database core comprises an acquisition module, a grouping module, an analysis module and a data module, wherein the acquisition module: the method supports multi-channel data import and collection; a grouping module: automatically clustering customer data through a big data algorithm; an analysis module: a multi-dimensional data report and a visual analysis interface; a data module: and (4) storage management of the bottom database.
Description
Technical Field
The invention relates to the technical field of calling systems, in particular to an AI intelligent client value management platform.
Background
The traditional call center services are labor-intensive, the demand of seat personnel is large, the loss is high, the service quality depends on the experience of the seat personnel, so that the cost of manpower is increased continuously for supporting the stable service, the wage and welfare cost is increased gradually along with the disappearance of population reducibility, the outdated data distribution, the marketing process and the result of traditional management tools such as Excel, Word and the like cannot be directly analyzed and controlled, the new service cannot be tested and reacted quickly, the communication resources lack professional call center operation and resource guarantee, the call cost is high, the manual conversion rate is low, and the data and the time are wasted.
Software and tools currently used by small and medium enterprises are lack of integration, a call center, communication resources, a BI (business information service) automatic outbound call system, a customer management system and the like are incompatible in system function and application, and multi-platform multi-environment processing is caused, so that the overall efficiency is influenced.
Therefore, the requirements for the product are divided into the following 5 aspects:
the number of the manual agents is effectively reduced by applying an AI technology in the call center system, so that the related investment of management cost, training cost, labor cost and the like is reduced, and the 'lightweight' of the call center is realized under the same service scale;
the automatic processing degree of the original service is effectively improved by combining the AI technology, and the capability of resisting the service fluctuation of an enterprise is greatly improved based on the characteristics of the voice robot, so that the operation stability is guaranteed;
the outbound capability of the AI voice robot is greatly superior to that of the traditional manual seat, and the service efficiency is effectively improved through a diversified human-computer interaction mode;
the system comprises a multi-dimensional customer portrait tool, a multi-element AI interactive scene and an intelligent distribution process management tool, and is used for assisting a user to accurately obtain customers, increasing conversion and improving the operational benefits;
the method has the advantages of flexible robot speech configuration and various robot voices, and can construct highly anthropomorphic interactive experience and provide satisfactory service experience for customers.
Disclosure of Invention
In order to solve the technical problems, the invention provides an AI intelligent customer value management platform integrating three main capabilities of a multi-channel entrance, an AI robot outbound and a big data analysis technology.
The AI intelligent customer value management platform of the invention comprises a database core, a collecting module, a grouping module, an analyzing module and a data module,
an acquisition module: support multi-channel data import and collection
Data are acquired in a multi-channel rich media mode, data types comprise basic data, a flat text file (Excel), audio data, video data and the like, all the data are acquired into a large data platform system, the platform can be adapted to support various types of data sources, structured data acquisition mainly comes from business system input, or is in butt joint with other peripheral systems through an API (application programming interface), or structured data is formed by text files and the like, the data platform constructs a database through ETL (Extract, Transform, Load), data acquisition is carried out on data required in the data platform from various channels every day or regularly according to a method established by a data warehouse, data adjustment is carried out according to various different channel data sources, and the platform extracts, cleans, combines and loads original data in the data acquisition process, the completeness of the data and the consistency of the data must be ensured in the process, when the business data volume is too large and the pressure of a Mysql data warehouse is not avoided to be too large, the business data can also be transferred to a database Hbase of a hadoop platform through a button.
In the unstructured data acquisition part, unstructured data such as video information, audio information, microblog real-time data, equipment data acquired by a sensor, data acquired by a mobile terminal and other streaming data are acquired through a sensor interface, video access equipment, a network crawler tool, a stream processing program and the like and are stored in the HDFS and the Hbase.
A grouping module: automatically clustering customer data through big data algorithm
The data platform utilizes machine learning algorithm technology based on the data that has been gathered, through characteristic classification form approximate clustering and grouping, the database training set comprises database Record (Record), every Record contains a plurality of fields or attribute (attribute), constitutes a feature vector, every Record of training set still has a specific label (ClassLabel) to correspond to it, this kind of label adopts the rule of thumb to be generated by the system automatically, the form of a specific sample can be sample vector: (v1, v 2.,. vn; c), where vi denotes field values and c denotes classes, the purpose of classification is to analyze the input data, and an accurate description or model is found for each class by the characteristics expressed by the data in the training set, and the class description thus generated is used to classify the future test data, although random variables are generated everywhere, the system can classify and define the future test data accordingly, so the system has a new label for each class in the database, and new data will be aggregated by approximate elements, so as to realize data learning, and new objective functions or rules will be derived under continuous learning, and each attribute set x is mapped to a predefined class label y.
The construction of the classification model in the data platform is divided into two stages of training and testing. Before constructing a model, a data set is required to be randomly divided into a training set and a testing set, in a training stage, the training data set is used, the model is constructed by analyzing database tuples described by attributes, the classification accuracy of the model is evaluated through data accumulation and spot inspection verification, and the optimization of the data tuples is continuously carried out, the classification algorithm of a platform refers to a KNN approximation algorithm (K-nearest neighbor) and is an example-based classification method and a non-parametric classification method, the classification is carried out by measuring the distance between different characteristic values, and the idea is as follows: if most of K most similar samples (namely, the most adjacent samples in the feature space) in the feature space belong to a certain class, the samples also belong to the class, wherein K is usually an integer not greater than 20, in the KNN algorithm, all selected neighbors are objects which are already correctly classified, and the method only determines the class to which the samples to be classified belong according to the class of the most adjacent sample or samples in the classification decision.
The KNN method is only related to a very small quantity of adjacent samples during class decision, so that the method can better avoid the imbalance problem of the samples, and in addition, since the KNN method mainly depends on the limited adjacent samples around, rather than on the method of distinguishing the class domain to determine the category, therefore, for the sample set to be classified with more cross or overlap of class domains, the KNN method is more suitable than other methods, the platform database aims at the problem of large calculation amount, the adopted solution method is to clip the known sample points in advance, remove the samples with little classification effect in advance, the samples are organized and sorted, grouped and layered, calculation is compressed within a small range close to the field of the test samples as much as possible, the distance calculation between a blind object and each sample in a training sample set is avoided, and the system is kept stable in the database operation process.
An analysis module: multidimensional data report and visual analysis interface
A visualization method and a visualization tool in data analysis provided by a data platform comprise a table, a histogram, a scatter diagram, a broken line diagram, a bar chart, a pie chart, an area diagram, a flow chart, a foam diagram and the like, a plurality of data series or combination of common diagrams like a time line, a Venn diagram, a data flow diagram, an entity relation diagram and the like, and a special analysis tool, such as a parallel coordinate used for drawing multi-dimensional individual data, a tree diagram used for visualizing a hierarchical structure, and a cone tree diagram used for layering of special data such as an organization body in a three-dimensional space, a semantic network and the like.
In addition, a data mining tool of the system is similar to Tableau and can support interactive and visual data analysis, a memory data engine is arranged in the data mining tool to accelerate visual processing, a data warehouse tool Hive based on Hadoop is utilized to structure the query and analyze cache information for the memory, the probability of Hadoop cluster delay is greatly reduced through the cache information, an interactive interaction mechanism is provided for users and big data application, the big data analysis tool of the platform can easily process ZB (terabyte) and PB (terabyte) data, the interactive visual cluster analysis method is the most direct method of a cluster mode, and visual multi-dimensional data is convenient for users to interactively analyze data and know the cluster structure.
Parallel coordinate and scatter plot matrices are generally used for analyzing data within ten dimensions, while star coordinates can process tens of dimensions, and cluster visualization based on star coordinates greatly improves data analysis availability by partially maintaining the position relationship by utilizing the performance of a potential mapping model.
A data module: storage management of underlying databases
The system platform jointly undertakes storage and management of structured data through relational databases MySQL and HBase databases of hadoop, realizes centralized storage and management of structured data and metadata by establishing a traditional data warehouse through MySQL, establishes a department and theme oriented data mart according to requirements, and divides a central data warehouse into three logic storage intervals: ODS (operational Data store), DW (Data Warehourse), DM (Data Mar): the ODS stores original data of each service system, wherein the original data comprise service data with the same structure as the original structure and service data subjected to preliminary arrangement; the DW area stores the sorted data and is a real data center of a big data analysis platform; the DM area stores integrated Data required by each application system (web application, BI, OLAP, Data Mining, etc.). Meanwhile, connection is established between the MySQL database and the HBase database, data exchange is carried out at regular time by using a button, and the two data warehouses share the big data application to provide data support, so that the aims of data sharing, pressure sharing and data backup are fulfilled.
In the aspect of unstructured data storage and management, because Mysql does not support storage of unstructured data, a data warehouse of a Hadoop platform of a big data application framework is used as supplement of a traditional data warehouse to realize storage and management of unstructured data and provide support for mass data query from a network, a Hadoop multi-functional component HDFS distributed file system stores big data files in a distributed mode, an Hbase extensible distributed column storage NoSQL database is used for storing structured and unstructured data, the platform meets the use requirements of personalized data of different users, the system arrangement provides three or more big data application modes, and the first support utilizes development languages such as java or C and the like to write programs to realize application of data in the Hadoop platform and the MySQL data warehouse; the second type is suitable for IBM-Cognos as an information sharing tool, Cognos is used as a diversified front-end analysis display tool, supports establishment of two models of DMR and OLAP, and provides various information sharing technologies such as online reports, OlAP analysis, instrument boards, scorecards, real-time query, Office integration, mobile APP and the like; and the third IBM-SPSS is used as a data mining tool, the SPSS supports a Hadoop platform and MySQL to build a mining model, is used for statistical analysis operation, data mining, prediction analysis and decision support tasks, and supports various types of statistical analysis and mining algorithms such as descriptive statistics, mean comparison, general linear model, correlation analysis, regression analysis, logarithmic linear model, cluster analysis, data simplification, survival analysis, time sequence analysis, multiple responses and the like.
The AI intelligent customer value management platform adopts a front-end and back-end separation framework as a whole, JavaScript language is used at the front end, an Ant Design of reach framework is adopted, Echarts is adopted at a report to realize front-end display, java development language is mainly adopted at the back end, and the front-end and back-end display platform comprises but not limited to springboot, mybatis, shiro, draud, tomcat, logback and easysexcel related components, and bottom-layer storage relates to mysql8.0, Redis 5.0.8, Apache Kafka 2.3.1, a Nacos configuration center, Ali cloud object storage and the like.
According to the AI intelligent client value management platform, the model obtained by classification learning is expressed in a classification rule form, a decision tree form or a mathematical formula form.
According to the AI intelligent customer value management platform, in the KNN algorithm, the distance between the objects is calculated to serve as a non-similarity index between the objects, so that the matching problem between the objects is avoided, and the KNN makes a decision according to the dominant class in the k objects instead of a single object class decision, which is the advantage of the KNN algorithm.
Drawings
FIG. 1 is an AI intelligence outbound intelligence assignment flow architecture diagram of the present invention;
FIG. 2 is a data management core diagram of the present invention;
FIG. 3 is a communication deployment architecture diagram of the present invention;
FIG. 4 is a core module architecture diagram of the present invention;
FIG. 5 is a business process architecture diagram of the present invention;
FIG. 6 is a database core block diagram of the present invention;
Detailed Description
The following detailed description of embodiments of the present invention is provided in connection with the accompanying drawings and examples. The following examples are intended to illustrate the invention but are not intended to limit the scope of the invention.
Example (b): as shown in fig. 1 to 6
The main technology is as follows: the system integrally adopts a front-end and rear-end separation framework, the front end uses JavaScript language, an Ant Design of React framework is adopted, the report forms realize front-end display by Echarts, the rear end mainly adopts java development language including but not limited to springboot, mybatis, shiro, druid, tomcat, logback and easysexcel related components, bottom storage relates to mysql8.0, Redis 5.0.8, Apache Kafka 2.3.1, Nacos configuration center, Aliskiu object storage and the like, and the stability and the safety of the whole project are integrally realized.
The acoustic model is a core technology of voice recognition, the voice acoustic model uses a combination of two deep learning network architectures, namely cnn + tdnnf, wherein cnn is used for enhancing voice feature extraction, and tdnnf is the latest acoustic modeling technology, language model research, big data deep learning and autonomous controllable voice transcription engine in the industry, so that speed response is improved.
Natural language understanding research, a robot system supports the abilities of intention, entity, variable, multi-round conversation assignment and the like, the customer intention is recognized in the form of grammar rules and natural language understanding (NLP), intention recognition accuracy is improved, training period and cost are reduced, when semantic similarity of sentences is calculated, a TensorFlow frame is used, multiple deep learning models such as MatchPyramid, DSSM, LSTM and BERT are trained for deeper semantic understanding, finally, the advantages of the models are integrated through analysis of results of different models, the own deep learning models are trained, reinforcement learning is introduced into the system, and accuracy of the models is continuously improved.
The BERT language representation model represents bidirectional encoder representation of a Transformer, and is different from other recent language representation models, the BERT aims to pre-train deep bidirectional representation by jointly adjusting contexts in all layers, therefore, the pre-trained BERT representation can be finely adjusted through an additional output layer, the method is suitable for building the most advanced model of a wide range of tasks, such as question-answering tasks and language reasoning, and does not need to make great architectural modification aiming at specific tasks, the bidirectional meaning represents that when processing a word, information of words in front of and behind the word can be considered, so that the semantics of the context is obtained, the training cost is reduced through datamation and model accumulation, and the machine learning efficiency is improved
Speech synthesis, also known as Text-to-Speech (Text-to-Speech) technology, can convert any Text information into standard smooth Speech in real time for reading, is equivalent to mounting an artificial mouth on a machine, relates to a plurality of subject technologies such as acoustics, linguistics, digital signal processing, computer science and the like, is a leading-edge technology in the field of Chinese information processing, and solves the main problem of how to convert the Text information into audible sound information, namely, how to make the machine speak like a person, and the research direction is the simulation degree, the smoothness and the synthesis budget time optimization.
Intelligent Speech Interaction (Intelligent Speech Interaction) is based on technologies such as Speech recognition, Speech synthesis and natural language understanding, and gives Intelligent man-machine Interaction experience of 'being able to listen, speak and understand you' to a product in various practical application scenes.
Difficult problems are as follows: the highly packaged interface is in butt joint with the call center, so that the system quickly supports the automatic outbound function of the AI on the premise of not influencing the whole software architecture and the deployment architecture, and the outbound of various scenes is completed.
The technical route includes that collected data are recycled through a multi-terminal (PC, applet, H5 and the like) collection and active reporting mechanism, the collected data are subjected to statistical analysis through various technical means such as direct processing, natural language processing, semantic analysis and the like, a complete analysis report is finally formed, feedback conditions of different types of questionnaires and different types of problems in the questionnaires are reflected, user groups are established for people needing special attention based on the collected data, and intelligent outbound return visits are performed on people needing important attention through a call center related technology, so that important attention of the people needing important attention and real-time monitoring of epidemic situation changes are achieved.
The implementation scheme is as follows:
1. data management core
Unifying a data management platform, managing all core data and tags without limiting expansion tags (self-definition), wherein the data management platform has the authority defined by total tag management, the data management platform and a business process platform interact data through an encryption interface (SSL), the business platform can read all tags of the data management platform, but the display content is subjected to authority management by the data management platform, the data management platform can import export data, if repeated tags are used, automatic tag filling can be carried out in a database collision, logs are generated, the data tags carry system tags such as business scenes (POS machines, education or other) and can be set, if the data in the database collision is carried out, the liveness is automatically updated, the data can define a structure with three-dimensional dimensions of encrypted and non-encrypted authority data, and the structure comprises follow-up records, circulation records, purchase records, tag records and the like, the data system has high availability including safety, stability and disaster tolerance, supports a dual-computer scheme, has an organization structure and management authority, can set a circulation mode including the daily request amount of subsystems, request limit and the like, has a high-availability analysis report form, and has automatic classification machine learning capability in the later stage.
2. Business subsystem design
Based on the system process design, the system tag content of the main database can be read, but data is acquired and displayed according to the authority of the subsystem, the foreground operation functions of a call center, such as click-to-dial, transfer, record checking and the like, the connection of a cloud end and a local IPPBX system is supported, number encryption is supported, data acquisition and subordinate distribution management are supported, the main process of electric marketing is supported, data from a data platform is directly converted to a company pool, a first-party telephone client is supported, a general client is supported, intention data and non-intention data are tracked to a subsystem public pool and are regularly recycled to a data management platform, the domain name access is supported by the subsystem, each branch company or channel client sandbox is deployed (namely independently deployed), a server supports at least 1000 users, the local deployment is supported by the subsystem, return visit reminding is supported, history follow-up and merchant are supported, the subsystem supports rapid deployment and maintenance.
3. IPPBX call center system design
The IPPBX system supports a scheme of N-to-N butt joint with a service system, the IPPBX supports basic soft switch services such as incoming call, outgoing call, screen flicking, recording, switching, busy display, idle display, click dialing, API (application programming interface), three-party call, voice quality inspection, call holding, call recovery and the like, the IPPBX supports cloud deployment and local deployment, common lines are supported, and a single machine supports 500 concurrence to the maximum.
4. Simple and clear visual center console for comprehensively mastering real-time service dynamic
The central console page is divided into modules of key index statistics, telephone traffic statistics, seat statistics and the like, which all display real-time data information of the same day, so that a user can conveniently and comprehensively master service dynamic, and the key index modules display call total amount, call receiving rate, manual work transfer amount, manual work transfer rate, unit forming amount, unit forming rate and other outbound and service data; the telephone traffic statistic module displays real-time call quantity, real-time call receiving rate, real-time switching number, real-time switching rate, robot concurrency quantity and the like; the agent counting module displays the telephone states of the extension numbers such as busy, idle, call, ringing and the like.
5. Custom client tag function to assist user in enriching management of client figures
The user-defined client tag function can be added or deleted according to the service needs of the user, the field type also provides multiple modes for input, relevant settings take effect in real time, and the timeliness of service adjustment is guaranteed.
The innovation points of the project are as follows: based on the saas cloud platform, the online account opening is fast to use, a plurality of API interface arrays are enriched, the fast butt joint of any peripheral system or core system is met, the call center, the BI, the CRM and the IPPBX are integrated in a whole core mode, the barrier of equipment and multiple platforms is broken through, and a set of system completes the penetration from a resource front end to a client terminal.
The method combines the AI intelligence and the traditional call center option type intelligent matching combined application mode, solves the problem of smooth transition of the AI intelligence in development, selects different man-machine combination modes according to different scenes and services, and fully, accurately and efficiently releases the advantage of the AI capability.
A delivery flow: docking communication lines → database deployment → system docking → account opening → voice robot configuration → business process setup → test → training and delivery.
The above description is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, several modifications and variations can be made without departing from the technical principle of the present invention, and these modifications and variations should also be regarded as the protection scope of the present invention.
Claims (4)
1. An AI intelligent customer value management platform is characterized in that a database core comprises an acquisition module, a grouping module, an analysis module and a data module,
an acquisition module: support multi-channel data import and collection
Data are acquired in a multi-channel rich media mode, data types comprise basic data, a flat text file (Excel), audio data, video data and the like, all the data are acquired into a large data platform system, the platform can be adapted to support various types of data sources, structured data acquisition mainly comes from business system input, or is in butt joint with other peripheral systems through an API (application programming interface), or structured data is formed by text files and the like, the data platform constructs a database through ETL (Extract, Transform, Load), data acquisition is carried out on data required in the data platform from various channels every day or regularly according to a method established by a data warehouse, data adjustment is carried out according to various different channel data sources, and the platform extracts, cleans, combines and loads original data in the data acquisition process, the completeness of the data and the consistency of the data must be ensured in the process, when the business data volume is too large and the pressure of a Mysql data warehouse is not avoided to be too large, the business data can also be transferred to a database Hbase of a hadoop platform through a button.
In the unstructured data acquisition part, unstructured data such as video information, audio information, microblog real-time data, equipment data acquired by a sensor, data acquired by a mobile terminal and other streaming data are acquired through a sensor interface, video access equipment, a network crawler tool, a stream processing program and the like and are stored in the HDFS and the Hbase.
A grouping module: automatically clustering customer data through big data algorithm
The data platform utilizes machine learning algorithm technology based on the data that has been gathered, through characteristic classification form approximate clustering and grouping, the database training set comprises database Record (Record), every Record contains a plurality of fields or attribute (attribute), constitutes a feature vector, every Record of training set still has a specific label (ClassLabel) to correspond to it, this kind of label adopts the rule of thumb to be generated by the system automatically, the form of a specific sample can be sample vector: (v1, v 2.,. vn; c), where vi denotes field values and c denotes classes, the purpose of classification is to analyze the input data, and an accurate description or model is found for each class by the characteristics expressed by the data in the training set, and the class description thus generated is used to classify the future test data, although random variables are generated everywhere, the system can classify and define the future test data accordingly, so the system has a new label for each class in the database, and new data will be aggregated by approximate elements, so as to realize data learning, and new objective functions or rules will be derived under continuous learning, and each attribute set x is mapped to a predefined class label y.
The construction of the classification model in the data platform is divided into two stages of training and testing. Before constructing a model, a data set is required to be randomly divided into a training set and a testing set, in a training stage, the training data set is used, the model is constructed by analyzing database tuples described by attributes, the classification accuracy of the model is evaluated through data accumulation and spot inspection verification, and the optimization of the data tuples is continuously carried out, the classification algorithm of a platform refers to a KNN approximation algorithm (K-nearest neighbor) and is an example-based classification method and a non-parametric classification method, the classification is carried out by measuring the distance between different characteristic values, and the idea is as follows: if most of K most similar samples (namely, the most adjacent samples in the feature space) in the feature space belong to a certain class, the samples also belong to the class, wherein K is usually an integer not greater than 20, in the KNN algorithm, all selected neighbors are objects which are already correctly classified, and the method only determines the class to which the samples to be classified belong according to the class of the most adjacent sample or samples in the classification decision.
The KNN method is only related to a very small quantity of adjacent samples during class decision, so that the method can better avoid the imbalance problem of the samples, and in addition, since the KNN method mainly depends on the limited adjacent samples around, rather than on the method of distinguishing the class domain to determine the category, therefore, for the sample set to be classified with more cross or overlap of class domains, the KNN method is more suitable than other methods, the platform database aims at the problem of large calculation amount, the adopted solution method is to clip the known sample points in advance, remove the samples with little classification effect in advance, the samples are organized and sorted, grouped and layered, calculation is compressed within a small range close to the field of the test samples as much as possible, the distance calculation between a blind object and each sample in a training sample set is avoided, and the system is kept stable in the database operation process.
An analysis module: multidimensional data report and visual analysis interface
A visualization method and a visualization tool in data analysis provided by a data platform comprise a table, a histogram, a scatter diagram, a broken line diagram, a bar chart, a pie chart, an area diagram, a flow chart, a foam diagram and the like, a plurality of data series or combination of common diagrams like a time line, a Venn diagram, a data flow diagram, an entity relation diagram and the like, and a special analysis tool, such as a parallel coordinate used for drawing multi-dimensional individual data, a tree diagram used for visualizing a hierarchical structure, and a cone tree diagram used for layering of special data such as an organization body in a three-dimensional space, a semantic network and the like.
In addition, a data mining tool of the system is similar to Tableau and can support interactive and visual data analysis, a memory data engine is arranged in the data mining tool to accelerate visual processing, a data warehouse tool Hive based on Hadoop is utilized to structure the query and analyze cache information for the memory, the probability of Hadoop cluster delay is greatly reduced through the cache information, an interactive interaction mechanism is provided for users and big data application, the big data analysis tool of the platform can easily process ZB (terabyte) and PB (terabyte) data, the interactive visual cluster analysis method is the most direct method of a cluster mode, and visual multi-dimensional data is convenient for users to interactively analyze data and know the cluster structure.
Parallel coordinate and scatter plot matrices are generally used for analyzing data within ten dimensions, while star coordinates can process tens of dimensions, and cluster visualization based on star coordinates greatly improves data analysis availability by partially maintaining the position relationship by utilizing the performance of a potential mapping model.
A data module: storage management of underlying databases
The system platform jointly undertakes storage and management of structured data through relational databases MySQL and HBase databases of hadoop, realizes centralized storage and management of structured data and metadata by establishing a traditional data warehouse through MySQL, establishes a department and theme oriented data mart according to requirements, and divides a central data warehouse into three logic storage intervals: ODS (operational Data store), DW (Data Warehourse), DM (Data Mar): the ODS stores original data of each service system, wherein the original data comprise service data with the same structure as the original structure and service data subjected to preliminary arrangement; the DW area stores the sorted data and is a real data center of a big data analysis platform; the DM area stores integrated Data required by each application system (web application, BI, OLAP, Data Mining, etc.). Meanwhile, connection is established between the MySQL database and the HBase database, data exchange is carried out at regular time by using a button, and the two data warehouses share the big data application to provide data support, so that the aims of data sharing, pressure sharing and data backup are fulfilled.
In the aspect of unstructured data storage and management, because Mysql does not support storage of unstructured data, a data warehouse of a Hadoop platform of a big data application framework is used as supplement of a traditional data warehouse to realize storage and management of unstructured data and provide support for mass data query from a network, a Hadoop multi-functional component HDFS distributed file system stores big data files in a distributed mode, an Hbase extensible distributed column storage NoSQL database is used for storing structured and unstructured data, the platform meets the use requirements of personalized data of different users, the system arrangement provides three or more big data application modes, and the first support utilizes development languages such as java or C and the like to write programs to realize application of data in the Hadoop platform and the MySQL data warehouse; the second type is suitable for IBM-Cognos as an information sharing tool, Cognos is used as a diversified front-end analysis display tool, supports establishment of two models of DMR and OLAP, and provides various information sharing technologies such as online reports, OlAP analysis, instrument boards, scorecards, real-time query, Office integration, mobile APP and the like; and the third IBM-SPSS is used as a data mining tool, the SPSS supports a Hadoop platform and MySQL to build a mining model, is used for statistical analysis operation, data mining, prediction analysis and decision support tasks, and supports various types of statistical analysis and mining algorithms such as descriptive statistics, mean comparison, general linear model, correlation analysis, regression analysis, logarithmic linear model, cluster analysis, data simplification, survival analysis, time sequence analysis, multiple responses and the like.
2. The AI smart customer value management platform of claim 1 wherein the system as a whole employs a front-end and back-end separation architecture, the front-end employs JavaScript language, the Ant Design of React framework is employed, the report forms employ Echarts to implement front-end presentation, the back-end employs primarily java development language including but not limited to springboot, mybatis, shiro, draid, tomcat, logback, easysecel related components, the underlying storage relates to mysql8.0, Redis 5.0.8, Apache Kafka 2.3.1, Nacos configuration center, Aliskiu object storage, and the like.
3. The AI smart client value management platform of claim 2, wherein the model obtained from classification learning is represented in the form of a classification rule, a decision tree, or a mathematical formula.
4. The AI intelligent customer value management platform of claim 3, wherein the KNN algorithm avoids matching problems between objects by calculating inter-object distances as indicators of dissimilarity between objects, and wherein the KNN algorithm takes advantage of the KNN algorithm by making decisions based on a dominant class among the k objects, rather than a single object class decision.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110021661.2A CN112685514A (en) | 2021-01-08 | 2021-01-08 | AI intelligent customer value management platform |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110021661.2A CN112685514A (en) | 2021-01-08 | 2021-01-08 | AI intelligent customer value management platform |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112685514A true CN112685514A (en) | 2021-04-20 |
Family
ID=75456352
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110021661.2A Pending CN112685514A (en) | 2021-01-08 | 2021-01-08 | AI intelligent customer value management platform |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112685514A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113506137A (en) * | 2021-07-14 | 2021-10-15 | 广州钛动科技有限公司 | E-mail marketing analysis method, system and equipment |
| CN113779761A (en) * | 2021-08-10 | 2021-12-10 | 南京莱斯信息技术股份有限公司 | Crowd defense and air defense organization data analysis system and method |
| CN117290429A (en) * | 2023-11-24 | 2023-12-26 | 山东焦易网数字科技股份有限公司 | Method for calling data system interface through natural language |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106709754A (en) * | 2016-11-25 | 2017-05-24 | 云南电网有限责任公司昆明供电局 | Power user grouping method based on text mining |
| CN109272155A (en) * | 2018-09-11 | 2019-01-25 | 郑州向心力通信技术股份有限公司 | A kind of corporate behavior analysis system based on big data |
| CN110163621A (en) * | 2018-02-10 | 2019-08-23 | 广州供电局有限公司 | A kind of electric power customer service big data DSS |
| CN110782318A (en) * | 2019-10-21 | 2020-02-11 | 五竹科技(天津)有限公司 | Marketing method and device based on audio interaction and storage medium |
| CN111475509A (en) * | 2020-04-03 | 2020-07-31 | 李俊宏 | Big data-based user portrait and multidimensional analysis system |
| CN111640040A (en) * | 2020-04-07 | 2020-09-08 | 国网新疆电力有限公司 | Power supply customer value evaluation method based on customer portrait technology and big data platform |
| CN111768850A (en) * | 2020-06-05 | 2020-10-13 | 上海森亿医疗科技有限公司 | Hospital data analysis method, hospital data analysis platform, device and medium |
| CN112070126A (en) * | 2020-08-21 | 2020-12-11 | 江西国云科技有限公司 | Internet of things data mining method |
| CN112184489A (en) * | 2020-09-30 | 2021-01-05 | 深圳供电局有限公司 | Power consumer grouping management system and method |
-
2021
- 2021-01-08 CN CN202110021661.2A patent/CN112685514A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106709754A (en) * | 2016-11-25 | 2017-05-24 | 云南电网有限责任公司昆明供电局 | Power user grouping method based on text mining |
| CN110163621A (en) * | 2018-02-10 | 2019-08-23 | 广州供电局有限公司 | A kind of electric power customer service big data DSS |
| CN109272155A (en) * | 2018-09-11 | 2019-01-25 | 郑州向心力通信技术股份有限公司 | A kind of corporate behavior analysis system based on big data |
| CN110782318A (en) * | 2019-10-21 | 2020-02-11 | 五竹科技(天津)有限公司 | Marketing method and device based on audio interaction and storage medium |
| CN111475509A (en) * | 2020-04-03 | 2020-07-31 | 李俊宏 | Big data-based user portrait and multidimensional analysis system |
| CN111640040A (en) * | 2020-04-07 | 2020-09-08 | 国网新疆电力有限公司 | Power supply customer value evaluation method based on customer portrait technology and big data platform |
| CN111768850A (en) * | 2020-06-05 | 2020-10-13 | 上海森亿医疗科技有限公司 | Hospital data analysis method, hospital data analysis platform, device and medium |
| CN112070126A (en) * | 2020-08-21 | 2020-12-11 | 江西国云科技有限公司 | Internet of things data mining method |
| CN112184489A (en) * | 2020-09-30 | 2021-01-05 | 深圳供电局有限公司 | Power consumer grouping management system and method |
Non-Patent Citations (2)
| Title |
|---|
| 姜巍;: "数据挖掘技术的分类算法分析研究", 电脑知识与技术, no. 01, 5 January 2009 (2009-01-05) * |
| 林赐云: "《吉林大学"十三五"本科规划教材 交通管理信息系统原理与应用》", 30 June 2019, 天津大学出版社, pages: 74 - 75 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113506137A (en) * | 2021-07-14 | 2021-10-15 | 广州钛动科技有限公司 | E-mail marketing analysis method, system and equipment |
| CN113779761A (en) * | 2021-08-10 | 2021-12-10 | 南京莱斯信息技术股份有限公司 | Crowd defense and air defense organization data analysis system and method |
| CN113779761B (en) * | 2021-08-10 | 2023-11-24 | 南京莱斯信息技术股份有限公司 | Crowd air defense organization data analysis system and method |
| CN117290429A (en) * | 2023-11-24 | 2023-12-26 | 山东焦易网数字科技股份有限公司 | Method for calling data system interface through natural language |
| CN117290429B (en) * | 2023-11-24 | 2024-02-20 | 山东焦易网数字科技股份有限公司 | Method for calling data system interface through natural language |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112685514A (en) | AI intelligent customer value management platform | |
| US20190286996A1 (en) | Human-machine interactive method and device based on artificial intelligence | |
| Burnap et al. | COSMOS: Towards an integrated and scalable service for analysing social media on demand | |
| CN109033387A (en) | A kind of Internet of Things search system, method and storage medium merging multi-source data | |
| CN109460737A (en) | A kind of multi-modal speech-emotion recognition method based on enhanced residual error neural network | |
| US20100161604A1 (en) | Apparatus and method for multimedia content based manipulation | |
| JP2020521210A (en) | Information processing method and terminal, computer storage medium | |
| CN109033277A (en) | Class brain system, method, equipment and storage medium based on machine learning | |
| CN111709575A (en) | Academic achievement prediction method based on C-LSTM | |
| CN111098312A (en) | Window government affairs service robot | |
| US20200125602A1 (en) | System and method for querying a data repository | |
| CN111813956A (en) | Knowledge graph construction method and device, and information penetration method and system | |
| US20190087828A1 (en) | Method, apparatus, and computer-readable media for customer interaction semantic annotation and analytics | |
| Karakus et al. | Call center performance evaluation using big data analytics | |
| CN103077207A (en) | Method and system for analyzing microblog happiness index | |
| Ahmed et al. | Sentiment analysis for smart cities: state of the art and opportunities | |
| CN112131203A (en) | Method and system for building data warehouse | |
| KR20190046062A (en) | Method and apparatus of dialog scenario database constructing for dialog system | |
| CN110334088A (en) | Educational data management system | |
| CN113656647B (en) | Intelligent operation and maintenance oriented engineering archive data management platform, method and system | |
| KR20180116104A (en) | Method and apparatus of dialog scenario database constructing for dialog system | |
| CN107977443A (en) | A kind of intelligent tutoring method and its system based on speech analysis | |
| CN111241288A (en) | Emergency sensing system of large centralized power customer service center and construction method | |
| CN115269862A (en) | Electric power question-answering and visualization system based on knowledge graph | |
| CN115080636A (en) | Big data analysis system based on network service |
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 |