CN103425776B - A kind of multi-user repository cooperation method - Google Patents
A kind of multi-user repository cooperation method Download PDFInfo
- Publication number
- CN103425776B CN103425776B CN201310355585.4A CN201310355585A CN103425776B CN 103425776 B CN103425776 B CN 103425776B CN 201310355585 A CN201310355585 A CN 201310355585A CN 103425776 B CN103425776 B CN 103425776B
- Authority
- CN
- China
- Prior art keywords
- subproblem
- knowledge
- knowledge base
- user
- confidence level
- 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
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention provides the method for a kind of multi-user repository cooperation, method is as follows: first problem is converted into question-based teaching, subproblem is assigned in corresponding knowledge base by recycling subproblem dispatching rules, and then solve this subproblem, utilize blackboard mechanism to carry out answer integration after subproblem solves to complete, finally under the associating of co-ordination principle storehouse and knowledge evaluation system, obtain final result and submit to user.It utilizes the real time correlation of multiple knowledge base to solve renewal of knowledge deficiency, and set up rigorous knowledge evaluation system and ensure the degree of accuracy and the reliability finally feeding back to user's answer, the evaluation score of the feedback real time modifying knowledge according to user optimizes the structure of knowledge and the knowledge degree of depth simultaneously；Utilize multiple knowledge base cooperation parallel processing to solve problem, improve problem-solving ability and efficiency.
Description
Technical field
The present invention relates to Internet technical field, provide a kind of multi-user repository cooperation method.
Background technology
People are endless for pursuit and the serious hope of knowledge, and the birth of internet and development make us enter the information age.In the face of immense problem and knowledge, how searching out the knowledge oneself wanted most becomes Times ' Demand with answer.
One of traditional acquisition knowledge mode is books, such as famous " Encyclopaedia Britannica ".But modern people are more likely to obtain oneself required knowledge in a network, more essential point is that renewal and the extension deficiency of such as " Encyclopaedia Britannica " this kind of books knowledge, especially relate to the field in some forward positions." Encyclopaedia Britannica " this kind of books belong to enclosed KBS, update slower.
For the KBS of the forms such as forum then due to the feature of forum so that knowledge may be faced when user searches required knowledge unprofessional, not comprehensively, the answer of even mistake.User oneself is then needed to go to screen, choose oneself required knowledge for the answer that the search engine such as Google, Baidu obtains.Lack a knowledge evaluation degree.
For the defect of prior art, the present invention proposes a kind of real-time system based on multi-user repository.Make full use of group intelligence making up the deficiency of machine learning, the problems such as the renewal of knowledge is not enough that utilize the real time correlation of multiple knowledge base to solve.Simultaneously set up rigorous knowledge evaluation system and ensure the degree of accuracy and the reliability finally feeding back to user's answer, carry out the evaluation score of real time modifying knowledge simultaneously according to the feedback of user.Thus optimize the structure of knowledge and the knowledge degree of depth.
Content of the invention
It is an object of the invention to provide one and have high efficiency, it is adaptable to extensive and express network, accuracy rate is high, it is ensured that ensure a kind of multi-user repository cooperation method that the degree of accuracy Detection accuracy of knowledge is high while the real-time of knowledge.
The present invention uses following skill scheme to achieve these goals:
A kind of multi-user repository cooperation method, it is characterised in that comprise the following steps:
The first step: the problem utilizing pretreatment module to submit user to pre-processes according to problem complexity etc., forms question-based teaching, and issues in the problem communal space.Wherein the problem communal space refers to may be used for showing solution problem, the virtual dynamic space of process, and blackboard mechanism common in Coordination Treatment technology can be used in actual applications to complete the work of the problem communal space；The problem first according to user's submission for the pretreatment module is as trunk, then using by the problem related with this problem and subtask as branch branch and leaf node, the most all subproblems layering of problem is enumerated, and starts the progressively question-based teaching of extension formation downwards from top.
Second step: problem is distributed to corresponding sub-knowledge base by subproblem dispatching rules storehouse by the problem of the problem communal space in succession so that sub-knowledge base can be with parallel processing.Subproblem dispatching rules storehouse refers to the node problems of question-based teaching scans in all of knowledge base retrieval, finds the knowledge base of corresponding subproblem and is distributed to these knowledge bases.Can assign according to the attribute of subproblem, feature etc. in being embodied as, the attribute of each subproblem all includes application domain, the wherein application domain in the attribute of the corresponding knowledge base of application domain, the feature of subproblem is then the feature interpretation for this subproblem, including the issuing time of subproblem, weight, the pointer etc. of father's problem.So when assigning, just can find corresponding sub-knowledge base territory according to the attribute of subproblem, then select corresponding knowledge base further according to some codomain (such as time priority, priority weights etc.) in its feature.
Result is submitted to the problem communal space by the 3rd step: after sub-knowledge base is retrieved and processed subproblem, if there is conflict rule between each knowledge base, then utilizes conflict rule storehouse to coordinate solution.Conflict rule storehouse is to solve priority when subproblem calls knowledge base knowledge and adopting series of rules set set by different knowledge base knowledge level problem.When being embodied as, can give the different subproblem node of weights and call the different priority of knowledge base and authority, priority can be realized by the higher simple algorithm of maximum weight priority；Knowledge is adopted degree and then can be represented by confidence level, and the bigger conclusion of final prioritizing selection confidence level is as the conclusion of subproblem；
4th step: the result of each subproblem obtaining inside the communal space is carried out being integrally formed answer, and answer is evaluated；
5th step: if the 4th step evaluation answer is unsatisfactory for previously given standard, artificial treatment will be carried out and knowledge base is fed back, thus the knowledge of more new knowledge base；
6th step: the problem answers obtaining is submitted to user.
In technique scheme, the described problem communal space is mainly used for showing problem, solves the virtual dynamic space of problem process, the purposes such as we can use blackboard coordination mechanism in force, and the problem that reaches is issued, process of problem solving is shown, problem is shared.In brief, blackboard can simply look at the intelligent agent realizing knowledge and information sharing, meets feature and the demand of the problem communal space.
In technique scheme, described subproblem dispatching rules storehouse is mainly used in carrying out knowledge base retrieval to subproblem and assigns, and first uses the clustering methods such as K-means to carry out classifying by subproblem, cluster according to the question attributes of subproblem when being embodied as.In classification, cluster process, employing depth-priority-searching method is carried out subproblem node traverses, it is to avoid omit subproblem node.Then use subproblem weights priority principle to carry out knowledge base assignment, so make subproblem can be assigned in order optimum knowledge base is processed, ensure that subproblem is processed simultaneously and be able to parallel processing, improve Resolving probiems efficiency.
In technique scheme, described conflict coordination rule base:
Identical conditions may meet multiple knowledge base, infer multiple relatively independent conclusion, thus cause knowledge to combine the expansion issues causing, in being then embodied as, we can be in the confidence level of the every rule of knowledge base knowledge acquisition stage definitions, thus select confidence level higher make inferences acquisition knowledge, if final result still can not meet confidence level or the appearance mistake of requirement, we will carry out abduction, use depth-priority-searching method to carry out again making inferences from big to small according to confidence level until meeting confidence level and requiring.And different knowledge bases call the stage we use the higher knowledge base that carries out of the bigger priority of weights of subproblem to call, when the conclusion that confidence level requires is met for final different knowledge bases, use the conclusion as this subproblem that confidence level is the highest；
The present invention is because using technique scheme, so possessing following beneficial effect:
First, high efficiency, it is adaptable to extensive and express network
One of design object of the present invention is to be deployed in the gateway of catenet and express network, such that it is able to utilize cluster network quickly to analyze and cooperative work, multiple knowledge base collaboration method that the present invention proposes is that to solve problem space too complicated too greatly thus cause inefficiencies.
2nd, the timely feedback module of the present invention can effectively solve redundancy and the old of knowledge, improve the degree of accuracy of knowledge further, inference machine engine (machine learning) powerful in addition and utilize the important evidence of the degree of accuracy that the wisdom of entity individual's cluster is all guarantee knowledge base knowledge.
3rd, real-time, it is adaptable to the scene of effectual requirement to knowledge feedback.
Brief description
Fig. 1 is the multiple knowledge base coordination model of the present invention.
Detailed description of the invention
The present invention uses to user knowledge base real time correlation, therefore can ensure the real-time of knowledge inside knowledge base.
The present invention it is critical only that the machine learning combination intelligent with entity personal set gunz, makes up the defect that traditional independent machine learning exists.Accompanying drawing 1 is shown in by whole scheme model, the foundation of user knowledge base and the basis that renewal is the present invention, and the cooperation of multi-user repository is then the key quickly solving challenge.
The invention mainly relates to problem pretreatment module, conflict coordination rule base, subproblem assign storehouse, result synthesis and the enforcement of evaluation module with cooperate.Key step is as follows:
The first step: the problem utilizing pretreatment module to submit user to pre-processes according to problem complexity etc., forms question-based teaching, and issues in the problem communal space.
Second step: problem is decomposed by the problem of the problem communal space by subproblem dispatching rules storehouse, and is in succession distributed to sub-knowledge base so that sub-knowledge base can be with parallel processing.
Result is submitted to the problem communal space by the 3rd step: after sub-knowledge base is retrieved and processed subproblem.If there is conflict rule between each knowledge base, then beneficially conflict rule storehouse coordinates solution.
4th step: each subproblem answer obtaining inside the communal space is carried out being integrally formed answer, and answer is evaluated.
5th step: if the 4th step evaluation answer is unsatisfactory for previously given standard, artificial treatment will be carried out and knowledge base is fed back, thus the knowledge of more new knowledge base.
6th step: the problem answers obtaining is submitted to user.
External pointer can be directly utilized carry out calling knowledge base when problem is fairly simple, carry out PROBLEM DECOMPOSITION even without dispatching rules storehouse, above-mentioned model mechanism then can be utilized to carry out decomposition to problem solve when problem is more complicated, each knowledge base can mean that a knowledge source, cover the special knowledge needed for specific area problem solving, relatively complete, an independent subproblem can be solved.
When resolution process is carried out to challenge, it is often necessary to the complexity according to problem itself carries out decomposing, dimensionality reduction operation etc., and subproblem to form relativity according to the primary and secondary of question attributes relatively low, the higher question-based teaching node of independence.The traversal of ad hoc rules can be carried out to question-based teaching, thus ensure to leave over subproblem, and after subproblem is processed completely, carry out the integration of result according to the inverse rule of traversal rule, so can ensure that reasonability and the correctness of result.
Leaf node in question-based teaching is all some relatively independent subproblems, give weights for each leaf node, each leaf node obtains a task chain from subproblem dispatching rules storehouse, task chain points to corresponding knowledge base, if knowledge base needs external pointer can be utilized to call when calling other relevant knowledge storehouses, and during solution problem, find that some knowledge base calls the external pointer that its external pointer frequently then can be set to fix, facilitate each knowledge base to call.
At this block of knowledge base, in order to obtain newer knowledge in time, we use real-time update mechanism, avoid the hysteresis quality of knowledge base knowledge as far as possible, then feed back to knowledge base according to evaluation module for those common sense knowledge storehouses so that knowledge base knowledge follows up the epoch in time.Knowledge connection problem for user knowledge base then can use cluster association, Attribute Association etc. according to actual conditions.
The answer that subproblem knowledge base solves to submit to is mainly carried out Knowledge Integration by result synthesis and evaluation module, and evaluates according to evaluation mechanism.The evaluation of knowledge base can use qualitative and quantitative mode to evaluate, and wherein qualitative analysis can be taked such as expert survey etc..Quantitative analysis can take multiattribute assessment method.Wherein evaluation index can select such as knowledge base content quality, Knowledge Base Techniques performance, the ease for use of knowledge base, knowledge base content quantity, the rate of people logging in of knowledge base and the ratio of error etc. of knowledge base according to actual conditions.
A simple application scenarios is set forth below: user have submitted a complicated computer cisco unity malfunction problem, then first problem is carried out pretreatment and obtain question-based teaching, his subproblem is likely to be software and causes operating system to be not normally functioning, the computer irregular working problem that hardware causes.But such subproblem is unduly complex, need to decompose further, till can not decomposing again.nullSuch as decomposing computer blue screen causes computer to be not normally functioning，Then we decompose computer blue screen code subproblem，Then emphasis retrieval is with regard to the knowledge base of computer blue screen code，And submit a question according to user and to obtain relevant information until searching the answer of blue screen code knowledge base，And the answer obtaining with other knowledge bases submits to jointly，Carry out arrangement integration finally according to the weights of subproblem answer and give evaluation module，Evaluation module provides best a series of answers according to customized good evaluation index，Next evaluate system feedback exactly and user obtain answer after solve, according to reality, the feedback that situation obtains，It should be noted that user feedback rank is above the priority of evaluation system，But in order to prevent the feedback of malicious user，We need to arrange user feedback certain authority，And the imparting of authority situation about in the past can use according to user and contribution degree，The attributes such as specialty degree give different authorities after carrying out overall merit.So both can increase the enthusiasm of user, it is also possible to utilize cluster wisdom to make up simple machine learning, the deficiency of reasoning.
Claims (1)
1. a multi-user repository cooperation method, it is characterised in that comprise the following steps:
The first step: utilize pretreatment module to pre-process the problem that user submits to according to problem complexity, forms question-based teaching, and issues in the problem communal space；
Second step: the problem in the problem communal space obtains subproblem by subproblem dispatching rules storehouse and problem is distributed to corresponding sub-knowledge base in succession；
Result is submitted to the problem communal space by the 3rd step: after sub-knowledge base is retrieved and processed subproblem, if there is conflict rule between each knowledge base, then utilizes conflict rule storehouse to coordinate solution；
4th step: the result of each subproblem obtaining inside the communal space is carried out being integrally formed answer, and answer is evaluated；
5th step: if the 4th step evaluation answer is unsatisfactory for previously given standard, artificial treatment will be carried out and knowledge base is fed back, thus the knowledge of more new knowledge base；
6th step: the problem answers obtaining is submitted to user；
The problem first according to user's submission for the pretreatment module is as trunk, then using by the problem related with this problem and subtask as branch branch and leaf node, the most all subproblems layering of problem is enumerated, and starts the progressively question-based teaching of extension formation downwards from top；
The problem communal space is for showing problem and the virtual dynamic space solving problem process, uses blackboard coordination mechanism, it is achieved problem is issued, process of problem solving is shown, problem is shared；
Subproblem dispatching rules storehouse is mainly used in carrying out knowledge base retrieval to subproblem and assigns,
First the question attributes according to subproblem uses clustering method to carry out classifying by subproblem, cluster；
In classification, cluster process, employing depth-priority-searching method is carried out subproblem node traverses, it is to avoid omit subproblem node；
Use subproblem weights priority principle to carry out knowledge base assignment, subproblem is assigned to all in order optimum knowledge base is processed；
Subproblem dispatching rules library collision co-ordination principle storehouse is as follows:
Giving the different subproblem node of weights and calling the different priority of knowledge base and authority, knowledge is adopted degree and then can be represented by confidence level, and the bigger conclusion of final prioritizing selection confidence level is as the conclusion of subproblem；
The confidence level of the every rule of knowledge base knowledge acquisition stage definitions, what selection confidence level was higher makes inferences acquisition knowledge, if final result still can not meet confidence level or the appearance mistake of requirement, carry out abduction, use depth-priority-searching method to carry out again making inferences from big to small according to confidence level until meeting confidence level and requiring；
And call the stage in different knowledge bases and use the higher knowledge base that carries out of the bigger priority of weights of subproblem to call, when the conclusion that confidence level requires is met for final different knowledge bases, use the conclusion as this subproblem that confidence level is the highest.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310355585.4A CN103425776B (en) | 2013-08-15 | 2013-08-15 | A kind of multi-user repository cooperation method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310355585.4A CN103425776B (en) | 2013-08-15 | 2013-08-15 | A kind of multi-user repository cooperation method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103425776A CN103425776A (en) | 2013-12-04 |
| CN103425776B true CN103425776B (en) | 2016-11-09 |
Family
ID=49650515
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310355585.4A Active CN103425776B (en) | 2013-08-15 | 2013-08-15 | A kind of multi-user repository cooperation method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103425776B (en) |
Families Citing this family (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104376365B (en) * | 2014-11-28 | 2018-01-09 | 国家电网公司 | A kind of building method in the information system operation rule storehouse based on association rule mining |
| CN105072195A (en) * | 2015-08-27 | 2015-11-18 | 阔地教育科技有限公司 | Information interaction server, transmitting terminal, processing terminal and information interaction system |
| CN105868179B (en) * | 2016-04-08 | 2018-07-13 | 上海智臻智能网络科技股份有限公司 | A kind of intelligent answer method and device |
| CN106155522B (en) * | 2016-06-29 | 2019-03-29 | 上海智臻智能网络科技股份有限公司 | Session data processing, knowledge base foundation, optimization, exchange method and device |
| CN107193872B (en) * | 2017-04-14 | 2021-04-23 | 深圳前海微众银行股份有限公司 | Question and answer data processing method and device |
| CN109214845A (en) * | 2017-07-03 | 2019-01-15 | 优甸网络科技（上海）有限公司 | The product introduction library of application product guiding |
| CN112818186A (en) * | 2021-01-22 | 2021-05-18 | 平安消费金融有限公司 | Information tracking method and device, electronic equipment and storage medium |
| CN113033156A (en) * | 2021-03-20 | 2021-06-25 | 广州快决测信息科技有限公司 | Logic tree based questionnaire processing method and device and storage medium |
| CN113033157B (en) * | 2021-03-20 | 2022-03-25 | 广州快决测信息科技有限公司 | Questionnaire generating method, device and storage medium |
| CN115277061A (en) * | 2022-06-13 | 2022-11-01 | 盈适慧众(上海)信息咨询合伙企业(有限合伙) | Network security service management system and method |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1500258A (en) * | 2000-12-22 | 2004-05-26 | ����̩�ϻ���ҽѧ��˾ | Multi-agent collaborative architecture for problem solving and tutoring |
| CN102902800A (en) * | 2012-10-12 | 2013-01-30 | 西安电子科技大学 | Agent-based intelligent meta search engine system |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101661570B (en) * | 2009-09-24 | 2012-01-25 | 河北工业大学 | Expert system for measurement of vacuum degree and life management of vacuum arc extinguish chamber and application method |
| CN101916328A (en) * | 2010-07-23 | 2010-12-15 | 东北电力大学 | Virtual electric power operation evaluating and analyzing method based on knowledge synergism |
-
2013
- 2013-08-15 CN CN201310355585.4A patent/CN103425776B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1500258A (en) * | 2000-12-22 | 2004-05-26 | ����̩�ϻ���ҽѧ��˾ | Multi-agent collaborative architecture for problem solving and tutoring |
| CN102902800A (en) * | 2012-10-12 | 2013-01-30 | 西安电子科技大学 | Agent-based intelligent meta search engine system |
Non-Patent Citations (1)
| Title |
|---|
| 基于多知识库协作的信息导航模型研究;李敏 等;《中国图书馆学报》;20060115(第1期);第51页，图1 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103425776A (en) | 2013-12-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103425776B (en) | A kind of multi-user repository cooperation method | |
| US11710070B2 (en) | Machine learned model framework for screening question generation | |
| Zhang et al. | An incremental CFS algorithm for clustering large data in industrial internet of things | |
| Lohr | The age of big data | |
| Krupitzer et al. | A survey on human machine interaction in industry 4.0 | |
| CN109564573A (en) | Platform from computer application metadata supports cluster | |
| Zhang et al. | Identifying network public opinion leaders based on markov logic networks | |
| Bryndin | Development of artificial intelligence by ensembles of virtual agents with mobile interaction | |
| WO2023163774A1 (en) | Individual treatment effect estimation under high-order interference in hypergraphs taking into account spillover effects | |
| Bogárdi-Mészöly et al. | Tag and topic recommendation systems | |
| CN106844934A (en) | Smart city planning and designing expert system and smart city planning and designing method | |
| DE CARLO et al. | Lean production and world class manufacturing: a comparative study of the two most important production strategies of recent times | |
| Kandul et al. | Explainable AI: A Review of the Empirical Literature | |
| US11687797B2 (en) | Real-time predictive knowledge pattern machine | |
| Gupta et al. | Bibliometric analysis of MOOC using bibliometrix package of R | |
| Bianchini et al. | The role of developers’ social relationships in improving service selection | |
| Maroun et al. | Ontology-based approach for cybersecurity recruitment | |
| Shikhalev et al. | Modelling of Conceptual Space of the “Regional Social Potential” Term on the Basis of Fuzzy Frames | |
| Khan | CAKE–Classifying, Associating and Knowledge DiscovEry-An Approach for Distributed Data Mining (DDM) Using PArallel Data Mining Agents (PADMAs) | |
| Eisenstadt et al. | Supporting Architectural Design Process with FLEA: A Distributed AI Methodology for Retrieval, Suggestion, Adaptation, and Explanation of Room Configurations | |
| US20230418837A1 (en) | System for Providing Data | |
| Köster et al. | An ontology for modelling human machine interaction in smart environments | |
| RU2707917C1 (en) | Method of searching for methods of resolving technical contradictions and a system based on a trained neural network for its implementation | |
| Wang et al. | A recommendation algorithm for collaborative denoising auto-encoders based on user preference diffusion | |
| Singh et al. | Latest Trends in Recommender Systems 2017 |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| CB03 | Change of inventor or designer information | ||
| CB03 | Change of inventor or designer information |
Inventor after: Zhang Xiaosong Inventor after: Chen Ruidong Inventor after: Niu Weina Inventor after: Wang Dong Inventor after: Chen Ting Inventor after: Pu Fulian Inventor after: Jiang Wei Inventor after: Liao Jun Inventor before: Chen Ting Inventor before: Zhang Xiaosong Inventor before: Chen Ruidong Inventor before: Niu Weina Inventor before: Wang Dong Inventor before: Pu Fulian Inventor before: Jiang Wei Inventor before: Liao Jun |
|
| COR | Change of bibliographic data |
Free format text: CORRECT: INVENTOR; FROM: CHEN TING ZHANG XIAOSONG CHEN RUIDONG NIU WEINA WANG DONG PU FULIAN JIANG WEI LIAO JUN TO: ZHANG XIAOSONG CHEN RUIDONG NIU WEINA WANG DONG CHEN TING PU FULIAN JIANG WEI LIAO JUN |
|
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |