CN114862081A - Case reasoning and artificial neural network-based aid decision-making method - Google Patents
Case reasoning and artificial neural network-based aid decision-making method Download PDFInfo
- Publication number
- CN114862081A CN114862081A CN202110154296.2A CN202110154296A CN114862081A CN 114862081 A CN114862081 A CN 114862081A CN 202110154296 A CN202110154296 A CN 202110154296A CN 114862081 A CN114862081 A CN 114862081A
- Authority
- CN
- China
- Prior art keywords
- case
- solution
- neural network
- data
- new
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/067—Enterprise or organisation modelling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Entrepreneurship & Innovation (AREA)
- Economics (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- General Health & Medical Sciences (AREA)
- Development Economics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Educational Administration (AREA)
- Health & Medical Sciences (AREA)
- Game Theory and Decision Science (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention relates to an assistant decision-making method based on case reasoning and an artificial neural network. Similar cases are matched in the model by inputting target cases according to a case retrieval algorithm based on a neural network, and the matched case scheme is modified according to rules or methods so as to adapt to new problems. And after the scheme is adjusted for many times, the final scheme is reserved to generate a new case, the newly generated case is compared with the case with the maximum similarity in the case base, if the maximum similarity threshold of the similar case is larger than the set corresponding similarity threshold, the newly generated new case replaces the returned source case and stores the source case into the case base, otherwise, the new case is directly stored in the case base, and the case base is updated. The invention makes full use of the CBR-ANN technology, reuses the prior case experience, helps the staff to quickly enter the problem, and improves the capability of making a solution and the customer satisfaction.
Description
Technical Field
The invention relates to the technical field of case reasoning technology and artificial neural network, in particular to an auxiliary decision-making method based on case reasoning and artificial neural network.
Background
With the rapid development of social economy, more and more small and medium-sized companies choose enterprise management consulting companies seeking specialities to make next development plans due to insufficient professional ability. The large amount of consulting case data is generally only accumulated in the company database and is not taken into consideration. The consultation of the target client only depends on the professional degree of the staff of the company and the experience of the staff of the company to give a decision. Even an experienced worker may have difficulty in giving professional management advice in a short time in conjunction with the operation situation of the target company and the current situation. With the rapid development of artificial intelligence, information technology with modern significance is combined into the management of consulting case data and the formulation of solutions, an intuitive and accurate case library is constructed, a complete case reference model is established, and powerful guarantee is provided for companies to formulate optimal solutions. The BP neural network algorithm is combined with the case-based reasoning technology, so that the intelligent BP neural network algorithm is intelligent, and a reference scheme can be accurately given.
The invention combines and applies case-Based reasoning technology CBR (case Based learning) and artificial neural network technology ANN (Artificial Intelligent network) to a scheme making auxiliary system so as to improve the quality of the solution designated by the company.
CBR is a further development of analogy-based learning, which uses past experience, methods and processes of problem solving for the solution of the current problem with reference to the role played by the previous cases in the learning and problem solving processes, wherein a case is an abstraction of a data set composed of four parts of consulting scheme information description, corresponding solution, scheme implementation trace records and service company operation information.
The CBR process can be regarded as a 4R (Retrieve, Reuse, Revise, Retain) loop process, i.e. a loop of four steps of similar case matching, case Reuse, case modification and adjustment, and case learning. When a client initiates consultation, consultation information and company information are integrated into a new case, the new case is input into a CBR system through case description, a case matched with the new case is found in a case library through a case retrieval algorithm based on a neural network, the matched scheme is modified according to rules or methods to adapt to a new problem, then the implementation condition of the tracking solution is continuously recorded, the scheme is adjusted in the process until the implementation of the scheme is stable, and the consultation scheme is summarized and arranged into the new case. For the generated new cases, the case descriptions are searched in a case library, and the case description with the maximum similarity is taken in the search result; and setting a corresponding similarity threshold, if the selected maximum similarity value is greater than the set corresponding similarity threshold, replacing the returned source case with the newly generated new case and storing the new case into the case library, otherwise, directly storing the new case into the case library to realize the updating of the case library.
The ANN is a machine learning technology which is widely applied in the aspects of classification and prediction at present, can process a large amount of data in parallel, is high in operation speed and short in time, needs to continuously adjust parameters in the operation executing process, adapts to a structure, and searches an optimal solution at a high speed, so that the aim of identification or operation can be finally achieved. The model is a three-layer BP neural network, existing data are used for training and simulating the network, the prediction performance of the neural network is exerted, effective reference data are provided for employees, and a good auxiliary effect is achieved for later scheme formulation.
Disclosure of Invention
The invention aims to integrate client consultation scheme data processed by an enterprise management consultation company into cases as a unit, establish a reference case base based on an artificial neural network, integrate consultation problems of clients and company operation information to form a new target case, input the target case into a model, search out similar cases through a case search algorithm based on the neural network, obtain a solution of a source case, modify the solution of the source case according to the similarity and difference of the source case and the target case to adapt to a new problem, continuously record and track the implementation condition of the solution, adjust the solution in the process until the proposal is stably implemented and arrange the consultation solution into the new case. Thereby assisting the staff in formulating consultation problem solutions.
In order to achieve the purpose, the invention provides the following technical scheme: an assistant decision-making system based on case reasoning technology comprises the following steps:
(1) acquiring financial information and a consulting scheme of a target client, integrating client consulting information and the general profile of a client company into a new target case, and inputting the target case;
(2) searching solutions of cases similar to the target case according to a case search algorithm based on a neural network, if matching is successful, obtaining one or more solutions, and modifying the solution of the source case according to the similarity and difference between the source case and the target case to adapt to the new problem because the case solution obtained by case matching cannot be directly used for solving the new problem;
(3) tracking the implementation condition of the solution, modifying and adjusting the solution until the solution runs smoothly, and reserving the modified final solution to generate a new case; if the matching is null, a new solution is automatically generated, the information of the case and the subsequent implementation condition are stored in a database, and a new case is generated;
(4) for the generated new cases, the case descriptions are searched in a case library, and the case description with the maximum similarity is taken in the search result; and setting a corresponding similarity threshold value x, if the selected maximum similarity value is greater than the set corresponding similarity threshold value, replacing the returned source case with the newly generated new case and storing the new case into the case base, otherwise, directly storing the new case into the case base to update the case base.
Drawings
FIG. 1 is a flow chart of an assistant decision method based on case-based reasoning and artificial neural network;
FIG. 2 is a diagram of a BP neural network model according to the present invention
FIG. 3 is a flow chart of an example of an aid decision according to the present invention;
FIG. 4 is a flowchart illustrating the management of newly created cases according to the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
As shown in fig. 1, the case-based reasoning and artificial neural network aided decision method provided by the embodiment of the present invention includes the following steps:
s101: collecting data such as detail information of each accepted consultation case, a case solution, a solution implementation tracking report, a service company general profile and the like stored in a database, and integrating the data into a data set;
s102: preprocessing the sorted data set, including data smoothing, time registration and the like, so as to improve the usability of data and samples; using Principal Component Analysis (PCA) to perform dimensionality reduction processing on the data, reducing data redundancy and keeping the most important aspect of the data;
s103: and storing the data subjected to the dimensionality reduction into a computer by taking a case as a unit, wherein the case is represented by case pair (case description, case solution), and the obtained cases form a case library. Determining a BP network model, establishing connection between a case base and the network model, learning, training and testing by using data in the case base, and finally establishing a proper BP neural network model;
s104: generating a new target case, inputting the target case into a case library for matching, successfully matching, finding out a solution of similar cases, and carrying out adaptive modification to obtain a final scheme; and if the matching is null, automatically generating a new solution, and obtaining a final solution in an auxiliary manner.
S105: for the generated new cases, the case descriptions are searched in a case library, and the case description with the maximum similarity is taken in the search result; and setting a corresponding similarity threshold, if the selected maximum similarity value is greater than the set corresponding similarity threshold, replacing the returned source case with the newly generated new case and storing the new case into the case library, otherwise, directly storing the new case into the case library to realize the updating of the case library.
As shown in fig. 2, the BP neural network model of the present invention is:
the BP neural network consists of an input layer, a hidden layer and an output layer, wherein the layers are connected through neurons, and each neuron in each layer is independent and is not connected. Each neuron input has a plurality of connection paths and only one output, each connection path corresponds to a weight coefficient, and the neural network learning and training process is mainly realized by adjusting the weight and a threshold of the network.
The transmission process of the BP neural network mainly comprises data forward propagation and error backward propagation. The BP neural network firstly carries out forward propagation, takes the data after standardization processing as the data of an input layer, transmits the data to a hidden layer through weight calculation, transmits the data to an output layer through the hidden layer, and enters an error backward propagation stage, namely a weight correction stage when an actual output result is inconsistent with an expected output result.
As shown in fig. 3, the assistant decision process of the present invention is:
(1) the staff integrates the consultation information and the accessory information of the target client into a target case, performs data preprocessing on the target case, and extracts the case problem characteristics;
(2) setting a similarity threshold value x of a target case, inputting the target case into a BP neural network model, searching and matching in the model, outputting case numbers, comparing similarity threshold values, and returning m previous cases with the similarity threshold value larger than the set similarity threshold value x;
(3) selecting a final case according to the fitness function and the algorithm stopping rule, and modifying the matched scheme by the staff according to the rule or the method so as to adapt to a new problem;
(4) and for the generated new cases, determining whether to store the cases in the case base or replace the cases in the case base according to rules, thereby realizing case learning. And the continuous updating of the case base is kept, and the maximum auxiliary effect is achieved.
As shown in fig. 4, the present invention relates to the subsequent processing of a new case generated after the counseling service is completed. After obtaining similar cases, we need to modify the matched scheme according to rules or methods to adapt to new problems. Tracking the implementation situation of the scheme, adjusting the scheme again, reserving the final scheme to generate a new case, retrieving the case description of the generated new case in a case library, and acquiring the case description with the maximum similarity in the retrieval result; and setting a corresponding similarity threshold, if the selected maximum similarity value is greater than the set corresponding similarity threshold, replacing the returned source case with the newly generated new case and storing the new case into the case library, otherwise, directly storing the new case into the case library to realize the updating of the case library.
Claims (5)
1. An assistant decision-making method based on case reasoning and artificial neural network comprises the following steps:
step one, collecting detail information, solutions, solution implementation tracking reports and service company profiles of each accepted consultation scheme stored in a database, and integrating the detail information, the solutions, the solution implementation tracking reports and the service company profiles into a data set;
secondly, preprocessing the well-regulated data set, including data smoothing, time registration and the like, and improving the usability of data and samples; using Principal Component Analysis (PCA) and Principal component analysis (Principal component analysis) to perform dimensionality reduction on data, reducing data redundancy and keeping the most important aspect of the data;
step three, storing the data after the dimensionality reduction processing into a computer by taking a case as a unit, wherein the case comprises a case description part and a case solution part, the case description part comprises detail information of a consultation scheme and a company profile, the case solution part comprises a solution scheme of the scheme and a solution implementation report, the obtained case forms a case base, and the case stored in the case base is a source case; optimizing the retrieval process of CBR by using a Back-Propagation Artificial Neural Network BP (Back-Propagation) and Back-Propagation Artificial Neural Network, firstly determining a BP Network model, establishing connection between a case base and the Network model, learning, training and testing by using a large amount of case data in the case base, and finally establishing a proper BP Neural Network model;
step four, sorting the information of the solution to be solved, generating a new target case, inputting the target case to perform retrieval matching in the model, returning a source case with similar capability to the target, and performing adaptive modification on the solution of the source case according to the similarity and difference between the source case and the target case to obtain a final solution; and for the generated new cases, determining whether to store the cases in the case base or replace the cases in the case base according to the principle of reserving the optimal solution, namely, only reserving the cases with the most effective solutions under the condition that the similarity reaches a preset value, thereby realizing case learning.
2. An aided decision making method based on case-based reasoning and artificial neural networks as claimed in claim 1, characterized in that: in the first step, a data set is collected, in which consultation plan information of the accepted clients, corresponding solutions, plan implementation tracking reports, and company profiles are collated into a combination.
3. An aided decision making method based on case-based reasoning and artificial neural networks as claimed in claim 1, characterized in that: in the second step, the preprocessing of the data is firstly to perform duplication and null removing operations on the collected data set to remove the data with messy codes, and finally to obtain a standard data set; and performing dimensionality reduction on the data by using PCA, wherein the main idea of the PCA is to map n-dimensional features onto k-dimensional features, the k-dimensional features are brand-new orthogonal features and are also called principal components, the k-dimensional features are reconstructed on the basis of the original n-dimensional features, namely, closely-related variables are changed into new variables as few as possible, the new variables are unrelated in pairs, namely, the data have no correlation in different orthogonal directions, so that a data set is easier to use, the calculation cost of a post-matching algorithm is reduced, and the processing speed of the valuable information of the sample is accelerated.
4. An aided decision making method based on case-based reasoning and artificial neural networks as claimed in claim 1, characterized in that: in the third step, after the dimensionality reduction processing, the consultation scheme information and the service company profile of the accepted client are used as case description, the solution and the solution implementation tracking report are used as case solution, the case description and the case solution form a case pair, the cases are represented by the case pair, and the obtained cases form a case library; the case comprises three parts, namely a description of the problem, a corresponding solution and a solution implementation effect, wherein the description of the problem and the corresponding solution are information which must be contained in the case description, the implementation effect of the solution is determined according to the requirements of the establishment of a case base, the case of the case base comprises two parts, namely a case description part and a case solution part, wherein the case description part comprises two parts, namely detail information of a consultation scheme and company outline, and the case solution part comprises two parts, namely a solution of the scheme and a solution implementation report;
determining a BP network model, connecting the BP network model with a case base, enabling the whole system to automatically acquire all case information for learning and training, constructing a reasonable case space, then utilizing a large amount of case data in the case base to automatically train and test a neural network, and finally acquiring an optimal BP neural network model, and in a CBR-ANN, training a neural network through a case matching algorithm based on the neural network to enable the neural network to output similar cases after case characteristics are input;
the BP neural network is composed of an input layer, a hidden layer and an input layer, each neuron input is provided with a plurality of connection paths, only one output is provided, each connection path corresponds to a weight coefficient, and the BP neural network is mainly realized by adjusting the weight and a threshold value of the network in the learning and training process of the neural network;
using a conjugate gradient BP algorithm, inIn error back propagation, let p k Direction of negative gradient change at kth iteration, q k The direction of the change of the conjugate gradient of the kth iteration is the weight correction amount delta omega k+1 And weight omega k+1 The method comprises the following steps:
ω k+1 =ω k +Δω k+1 +η k q k
suppose that there are L sets of training data with m inputs 1 Dimension vectorLet w i The weight ratio of each input quantity is adjusted for the connection weight value, and the output is m 3 The training output of the k-th group of data of the dimension vector, neural network isy k As a result of expected or accurate results; initial time, q 0 =-p 0 Each iteration is preceded by a negative gradient direction p k Searching, and then searching along the direction of the conjugate gradient, wherein the direction of the conjugate gradient is a linear combination of the current direction of the conjugate gradient and the previous direction of the conjugate gradient, namely:
η k =E/S
in the formula:
5. an aided decision making method based on case-based reasoning and artificial neural networks as claimed in claim 1, characterized in that: in the fourth step, a target customer consultation scheme and customer information are obtained, a target case is integrated and generated, case description is used as characteristics of the target case and input into a network model, retrieval is carried out in the model to obtain a similar source case, the solution of the source case obtained by retrieval cannot be directly used for solving a new problem, and the solution of the source case needs to be modified according to the similarity and difference between the source case and the target case so as to adapt to the new problem; tracking the implementation situation of the scheme, adjusting the scheme again, reserving the final scheme to generate a new case, describing the case of the generated new case in a case library for retrieval, and acquiring a source case with the maximum similarity from a retrieval result; and according to the set corresponding similarity threshold value x, if the selected maximum similarity value is greater than the set corresponding similarity threshold value x, replacing the returned source case by the newly generated new case and storing the new case into the case base, otherwise, directly storing the new case into the case base, thereby realizing the updating of the case base.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110154296.2A CN114862081A (en) | 2021-02-04 | 2021-02-04 | Case reasoning and artificial neural network-based aid decision-making method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110154296.2A CN114862081A (en) | 2021-02-04 | 2021-02-04 | Case reasoning and artificial neural network-based aid decision-making method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN114862081A true CN114862081A (en) | 2022-08-05 |
Family
ID=82622750
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110154296.2A Pending CN114862081A (en) | 2021-02-04 | 2021-02-04 | Case reasoning and artificial neural network-based aid decision-making method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114862081A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116702334A (en) * | 2023-08-04 | 2023-09-05 | 中国人民解放军国防科技大学 | Sparse storage method for overall design case of solid engine |
CN117350288A (en) * | 2023-12-01 | 2024-01-05 | 浙商银行股份有限公司 | Case matching-based network security operation auxiliary decision-making method, system and device |
-
2021
- 2021-02-04 CN CN202110154296.2A patent/CN114862081A/en active Pending
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116702334A (en) * | 2023-08-04 | 2023-09-05 | 中国人民解放军国防科技大学 | Sparse storage method for overall design case of solid engine |
CN116702334B (en) * | 2023-08-04 | 2023-10-20 | 中国人民解放军国防科技大学 | Sparse storage method for overall design case of solid engine |
CN117350288A (en) * | 2023-12-01 | 2024-01-05 | 浙商银行股份有限公司 | Case matching-based network security operation auxiliary decision-making method, system and device |
CN117350288B (en) * | 2023-12-01 | 2024-05-03 | 浙商银行股份有限公司 | Case matching-based network security operation auxiliary decision-making method, system and device |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Ren et al. | Sequential recommendation with self-attentive multi-adversarial network | |
CN114862081A (en) | Case reasoning and artificial neural network-based aid decision-making method | |
CN109359302B (en) | Optimization method of domain word vectors and fusion ordering method based on optimization method | |
CN109902192B (en) | Remote sensing image retrieval method, system, equipment and medium based on unsupervised depth regression | |
CN117290462B (en) | Intelligent decision system and method for large data model | |
CN103886030A (en) | Cost-sensitive decision-making tree based physical information fusion system data classification method | |
CN116644170A (en) | Reply text generation method, device, communication equipment and storage medium | |
CN117668205B (en) | Smart logistics customer service processing method, system, equipment and storage medium | |
CN113158555A (en) | Heavy gas turbine control system analog input module BIT design method based on expert system and random forest classifier | |
Naing et al. | Feature Selection for Customer Churn Prediction: A Review on the Methods & Techniques applied in the Telecom Industry | |
Lavingia et al. | Predicting stock market trends using random forest: A comparative analysis | |
Mittal | An approach towards enhancement of classification accuracy rate using efficient pruning methods with associative classifiers | |
Liu et al. | A provenance meta learning framework for missing data handling methods selection | |
CN114116692B (en) | Mask and bidirectional model-based missing POI track completion method | |
CN112949599B (en) | Candidate content pushing method based on big data | |
CN112749530B (en) | Text encoding method, apparatus, device and computer readable storage medium | |
CN114240318A (en) | Target object oriented information processing method and device and computer equipment | |
CN113254468A (en) | Fault query and reasoning method for certain type of equipment | |
Yahia et al. | K-nearest neighbor and C4. 5 algorithms as data mining methods: advantages and difficulties | |
Bengoetxea et al. | Solving graph matching with EDAs using a permutation-based representation | |
Ferreira et al. | A hybrid intelligent system approach for improving the prediction of real world time series | |
JP3212695B2 (en) | Knowledge acquisition device for knowledge base system and knowledge correction device | |
JP2021081930A (en) | Learning device, information classification device, and program | |
Djeddi et al. | Introducing artificial neural network in ontologies alignment process | |
CN118036756B (en) | Method, device, computer equipment and storage medium for large model multi-round dialogue |
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 |