CN110489749A - Intelligent Office-Automation System Work Flow Optimizing - Google Patents
Intelligent Office-Automation System Work Flow Optimizing Download PDFInfo
- Publication number
- CN110489749A CN110489749A CN201910723500.0A CN201910723500A CN110489749A CN 110489749 A CN110489749 A CN 110489749A CN 201910723500 A CN201910723500 A CN 201910723500A CN 110489749 A CN110489749 A CN 110489749A
- Authority
- CN
- China
- Prior art keywords
- term vector
- official document
- document
- layer
- word
- 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.)
- Granted
Links
Classifications
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- 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
- G06Q10/10—Office automation; Time management
- G06Q10/103—Workflow collaboration or project management
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Entrepreneurship & Innovation (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Economics (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention discloses a kind of intelligent Office-Automation System Work Flow Optimizing method, and the intelligent Office-Automation System operation flow mainly includes the building of the office system environment based on BPM workflow engine, the building in official document Word2vec term vector library, the regularization processing in term vector library, the extraction of official document feature set and the document information identification based on convolutional neural networks.Office system environment based on BPM workflow engine realizes the lasting upgrading optimization of information transmitting, data synchronization, business monitoring and business events flow path.The mapping of the system of the Environmental Support realistic model management business and workflow, comprehensively precisely defines the operation content of each task and operational staff, converts unified, automatic, standard computer system operation for Business Process Management.The Work Flow Optimizing based on convolutional neural networks is furthermore achieved in the building that the present invention also passes through official document Word2vec term vector library, cooperates with the various system resources of enterprises, reduces time cost, improve the response efficiency of business process system.
Description
Technical field
The present invention relates to the document information intelligent identification technology in the office automation system, realize that operation flow pushes away automatically
It recommends, relates generally to the building in official document Word2vec term vector library and the document information identification based on convolutional neural networks.
Background technique
Recently as the heating of big data and artificial intelligence concept, more and more enterprises start to pay attention to containing in data
In commercial value, and increasingly mature with data mining technology, machine learning is in government, finance, communication, medical treatment, zero
Sell etc. in multiple industries using more and more extensive and deeply, based on specific transactions target problem and historical data building data point
Analyse model, the model managements behaviors such as implementation model storage, model verifying, model application deployment and model performance monitoring, it has also become
Important link in business data mining process.To cope with continually changing market environment and continually changing customer demand,
How to establish it is efficient, repeatable, can iteratively faster, and the model management process that flexibility ratio is high is that enterprise promotes the analysis of its data
Ability and by its ability and resource conversion be value one of key subject, there is very strong research and practice significance.This patent
The application model of Data Analysis Model management and workflow composing is explored based on this direction.
Conventional message-driven or event driven Workflow system can preferably support the operation of office system, but
It is that can not keep up with the rapid development of artificial intelligence.It realizes and the combination of document information and historical data analysis and mentions
For intelligent Service, it is necessary to be included in data driven technique in former Workflow system.In addition, Chinese government's office is mainly with official document
For foundation, so the importance of the intelligent Service based on document is especially prominent.How document cooperateed with into processing technique and semanteme
Processing technique is used for office system, realizes document information intelligent recognition and provides intelligent Service for user, needs solve
Major issue.
Although existing system changes original document treatment mode, do not change the process flow of official document,
I.e., it has no advanced management method increase is come in.Product more commonly used at present is made a general survey of, what they were capable of providing
Function is substantially the processing around official document, and no matter additional function for depth or range far can not all enable us
It is satisfied.Its basic reason is exactly that these systems only merely handle official document, and oneself processing official document is not used, that is,
Say, do not accomplish accumulation of knowledge, can not oneself has according to these information be that enterprise needs the service of making a policy.Official document system exists
Official document does not have automatic official document automating sorting function when distributing, and the requirement for user is relatively high, this is need urgently to be resolved
It asks.In addition, official document caused by enterprise increases sharply with the rapid expansion of information technology, data volume is huge in process log how
More efficiently effectively handled official business demand just urgently to be resolved at Liao Ge company using these data.Therefore, under privately owned cloud environment
The characteristics of magnanimity, dispersion, dynamic change is presented in information data, needs to carry out the Business Stream under privately owned cloud environment based on data-driven
Cheng Youhua key technology research.
Summary of the invention
Intelligent Office-Automation System operation flow recommends the method for being based primarily upon data-driven intelligently to know document information
Not, the function of realization mainly has: being drawn based on BPM (Business process management, Business Process Management) workflow
The building of the office system environment held up, the building in official document Word2vec term vector library, the regularization processing in term vector library, official document are special
The extraction of collection and based on convolutional neural networks document information identification, as shown in Figure 1.
Intelligent Office-Automation System use the office system environment based on BPM workflow engine, realize information transmitting,
The lasting upgrading optimization of data synchronization, business monitoring and business events flow path.The Environmental Support realistic model management business and work
Make the system flowed mapping, comprehensively the operation content of each task and operational staff is precisely defined, by operation flow pipe
Reason is converted into unified, automatic, standard computer system operation.This patent incorporates deep learning in this context simultaneously
Frame as auxiliary support, be furthermore achieved by the building in official document Word2vec term vector library based on convolutional neural networks
Work Flow Optimizing, cooperate with the various system resources of enterprises, reduce time cost, improve the sound of business process system
Answer efficiency.
The present invention has the advantages that
1. the operation flow recommender system is using the OA based on BPM workflow engine, (Office Automation, is done
Public affairs automation) system, conventional operation stream engine is compared, the system information conformability is strong, and the circulation data between document pass in real time
It passs, and all kinds of IT systems and resource can be integrated, unified flow services are provided for other application, be fusion much information service
Operation flow examination & approval management platform.In addition, the configurability of process height can greatly reduce exploitation amount, quick, low cost pipe
Simple or complicated all kinds of operation flow lists are managed, meanwhile, the unified management of artificial process and automatic flow can adapt to process
Fast-changing needs.
2. the technology in official document Word2vec term vector library is by neural network to context semanteme and context and target
Between relationship modeled.This method proposes the form for converting word to vector for complicated document information, effectively solves
It has determined the problem, dimension can be reduced and capture position related information of some word in official document.
3. Work Flow Optimizing system is on the basis of Word2vec term vector library technology, introducing TF-IDF, (word frequency-is inverse
Text frequency) algorithm calculates the weight of each term vector in the text, and it is public for assessing the portion in a words or corpus
The significance level of text.This for BPM workflow engine provide as degree of correlation between file and user query measurement or
Grading.
4. Work Flow Optimizing system realizes the intelligent recognition of document information based on convolutional neural networks, difference has been used
Convolution kernel convolution is carried out to official document matrix, the width of convolution kernel is equal to the length of term vector, then uses max-pooling
The vector that (maximum value sampling) extracts each convolution kernel operates, the last corresponding number of each convolution kernel, this
A little convolution kernels are stitched together, and have just obtained the sentence vector for characterizing the official document, have finally obtained business by each section Weight
The recommendation results of process improve the accuracy of the efficiency and operation flow of office worker.
Detailed description of the invention
Fig. 1 is the Work Flow Optimizing system function frame based on BPM workflow engine
Fig. 2 is BPM workflow engine frame
Fig. 3 is operation flow message flow exemplary diagram
Fig. 4 is the mapping relations of official document and term vector based on Word2vec
Fig. 5 is feature extraction and the weight distribution flow chart of official document
Fig. 6 is the official document classification model of convolutional neural networks
Fig. 7 is operation flow intelligent recommendation system interface
Specific embodiment
The Work Flow Optimizing of intelligent Office-Automation System provided by the invention is carried out specifically with reference to the accompanying drawing
It is bright:
1. the OA office system based on BPM workflow engine
BPM is the management based on operation flow itself, the design of comprehensive support process, execute, management, association, optimization it is each
A aspect full-automation Collaboration.In order to which service procedure and artificial process are all integrated into BPMN (BPM Network),
A kind of architecture design of complexity is needed, as shown in Fig. 2, the BPM engine for incorporating BPEL4People is the real core of whole system
BPMN2 design test deployment process can be used to wherein in the heart, domain expert, and management console can be carried out each of process flow operation
Kind management, BPM engine are not only responsible for sending task to human task system, be also responsible for each REST
(Representational State Transfer) endpoint service interactive correspondence.Wherein BPM message flow acts on BPMN cooperation
Figure, collaboration diagram show how two or more no central controlled processes interact in a synchronous manner.Message flow is table
The mode how process individually controlled up to two is in communication with each other and cooperate, activity or event in a pond can be to another
Message is initiated in pond, and message flow is depicted as a dotted line, and hollow circles indicate the source of message, and empty arrow indicates message expiration, disappears
Breath stream example is as shown in Figure 3.Each process is included in the pond Pool of oneself, and pond is often marked with participant's title.Face
Various resources and service can be driven to the process of service management, realizes a variety of inquiries and statistical analysis.The office system will be believed
Breath exchange, knowledge, notice bulletin etc. are unified on process portal, improve the tissue in process implementation procedure and team collaboration, mistake
Journey and result can all analyze retrospect.
2. the building in official document Word2vec term vector library and the extraction of official document feature set
Word2vec is the correlation model that a group is used to generate term vector.These models are double-deck neural network, are passed through
Train construction term vector library again.After training is completed, word2vec model can be used to map each word to a vector, this to
Amount can be used to indicate the relationship between word and word, and specific mapping process is as shown in Figure 4.
The training process of Word2vec is as follows:
1) word after word segmentation processing is indicated by the way of only heat type coding.
2) term vector obtained in the previous step is multiplied into projection matrix and obtains the input of hidden layer, then obtained by activation primitive hidden
Hide the output of layer.But it calculates to generally use for simplicity and the output of input layer directly summation is obtained into the input of hidden layer.
3) output layer is a Hofman tree, and wherein leaf node corresponds to the word in vocabulary, and non-leaf nodes is equivalent to
Parameter of the hidden layer to output layer.Pass through the weight between the available input layer of training pattern and hidden layer.
4) word that initial only heat type encodes is indicated and the multiplied by weight between input layer and hidden layer obtains word
Term vector indicate.
3. the feature extraction and weight distribution of official document
This submodule is first to official document diCarry out word segmentation processing Wi=[w1,w2,…,wn], n be word number, then further according to
Text after participle is replaced with low-dimensional numerical value vector by Word2vec term vector library For wi
Term vector,K is the dimension of term vector, makes official document expressing from the intractable high latitude of neural network
High sparse traditional data, the continuous dense matrix data for becoming similar image indicate.This official document expressing method avoids tradition
The cumbersome work that manual features select in machine learning text classification algorithm, allows document information to obtain maximum reservation,
And this patent uses modified TF-IDF algorithm and carries out term vector weight calculation, and specific process flow is as shown in Figure 5.
4. the document information based on convolutional neural networks identifies
Convolutional neural networks are generally made of input layer, multiple convolutional layers, pond layer, full articulamentum and softmax layers, are used
It is as shown in Figure 6 in the convolutional neural networks model of document information identification.By multiple and different convolution kernel ω, ω in one convolutional layer
∈Rhk, h is the height of convolution kernel, and k is the Spatial Dimension of term vector, and convolution kernel is every to pass through a text with 1 slide downward of step-length
Convolution algorithm is carried out when the window of vector h*k, generates a new characteristic value.Wi:i+hThe sequence of terms for being h+1 for a length
(Wi,Wi+1,…,Wi+h), ω is convolution kernel matrix weight parameter, and b is bias term, and b ∈ R, operator () is convolutional calculation, f
For activation primitive.One convolution kernel obtains a characteristic pattern c=(c after handling text vector1,c2,…,cn-h+1), n is official document
The number of middle word.Pond layer is extracted using feature of the 1-max-pooling to characteristic pattern, cm=max { c }, passes through pond
After changing layer processing, the text of different length all becomes the feature of equal length.The input of full articulamentum is the feature of pond layer
Output, inputs and isP is the type of convolution kernel, and q is every kind of convolution kernel
Number, output layer carries out kind judging using softmax function, to reach operation flow intelligent recommendation, implements boundary
Face is as shown in Figure 7.
Claims (4)
1. intelligent Office-Automation System Work Flow Optimizing, it is characterised in that: the office system ring based on BPM workflow engine
The building of border, the building in official document Word2vec term vector library, the regularization processing in term vector library, the extraction of official document feature set and base
It is identified in the document information of convolutional neural networks.Office system environment based on BPM workflow engine, realize information transmitting,
The lasting upgrading optimization of data synchronization, business monitoring and business events flow path；The building in official document Word2vec term vector library will be gone through
History DOC DATA is mapped to term vector, efficiently solves the problems, such as the expression of official document complex information；The regularization in term vector library is handled
It ensure that the consistency and validity of document information format, with the extraction of official document feature set convenient for carrying out official document using neural network
Information processing；Document information identification based on convolutional neural networks carries out convolution weighting point to the matrix that official document term vector is constituted
The recommendation results of operation flow can be obtained in class, improve the accuracy of the efficiency and operation flow of office worker.
2. intelligent Office-Automation System Work Flow Optimizing according to claim 1, it is characterised in that: the official document
The relationship of the neural network to context semanteme and between context and target that be constructed by Word2vec term vector library carries out
Modeling.Word in the present invention by official document after word segmentation processing is indicated by the way of one-hot coding；It will be obtained in the previous step
Term vector multiplies projection matrix and obtains the input of hidden layer, then the output of input layer directly summation is obtained the input of hidden layer；It is defeated
Layer is a Hofman tree out, and wherein leaf node is exactly the word in corresponding vocabulary, and non-leaf nodes is equivalent to hidden layer and arrives
The parameter of output layer passes through the weight between the available input layer of training pattern and hidden layer；Initial only heat type is encoded
Word indicate and multiplied by weight between input layer and hidden layer is to obtain the term vector of word to indicate.This method is for complexity
Document information proposes the form for converting word to vector, efficiently solves the problems, such as this, can reduce dimension and capture
Position related information of some word in official document.
3. intelligent Office-Automation System Work Flow Optimizing according to claim 1, it is characterised in that: term vector library
The extraction of regularization processing and official document feature set is utilized TF-IDF algorithm and has carried out at distribution to term vector weight in the present invention
Reason, first to official document diCarry out word segmentation processing Wi=[w1,w2,…,wn], n is word number, then further according to Word2vec term vector
Text after participle is replaced with low-dimensional numerical value vector by library For wiTerm vector,K is the dimension of term vector, makes official document expressing from the high sparse biography of the intractable high latitude of neural network
System data, the continuous dense matrix data for becoming similar image indicate.This official document expressing method avoids conventional machines study
The cumbersome work that manual features select in text classification algorithm, allows document information to obtain maximum reservation.
4. intelligent Office-Automation System Work Flow Optimizing according to claim 1, it is characterised in that: based on convolution mind
Document information identification through network, at input layer, multiple convolutional layers, pond layer, full articulamentum and softmax layers of information
Reason.By multiple and different convolution kernel ω, ω ∈ R in one convolutional layerhk, h is the height of convolution kernel, and k is the space dimension of term vector
Degree, for convolution kernel with 1 slide downward of step-length, when every window by a text vector h*k, carries out convolution algorithm, generate one it is new
Characteristic value.Wi:i+hSequence of terms (the W for being h+1 for a lengthi,Wi+1,…,Wi+h), ω is convolution kernel matrix weight parameter,
B is bias term, and b ∈ R, operator () is convolutional calculation, and f is activation primitive.One convolution kernel obtains after handling text vector
To a characteristic pattern c=(c1,c2,…,cn-h+1), n is the number of word in official document.Pond layer is using 1-max-pooling to feature
The feature of figure extracts, cm=max { c }, after being handled by pond layer, the text of different length all becomes the spy of equal length
Sign.The input of full articulamentum is the feature output of pond layer, inputs and is
P is the type of convolution kernel, and q is the number of every kind of convolution kernel, and output layer carries out kind judging using softmax function, to reach
To operation flow intelligent recommendation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910723500.0A CN110489749B (en) | 2019-08-07 | 2019-08-07 | Business process optimization method of intelligent office automation system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910723500.0A CN110489749B (en) | 2019-08-07 | 2019-08-07 | Business process optimization method of intelligent office automation system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110489749A true CN110489749A (en) | 2019-11-22 |
| CN110489749B CN110489749B (en) | 2021-05-18 |
Family
ID=68550041
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910723500.0A Active CN110489749B (en) | 2019-08-07 | 2019-08-07 | Business process optimization method of intelligent office automation system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110489749B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111782811A (en) * | 2020-07-03 | 2020-10-16 | 湖南大学 | E-government affair sensitive text detection method based on convolutional neural network and support vector machine |
| CN112712177A (en) * | 2020-12-29 | 2021-04-27 | 上海永骁智能技术有限公司 | Knowledge engineering method and device based on cooperative processing |
| CN114331226A (en) * | 2022-03-08 | 2022-04-12 | 天津联创科技发展有限公司 | Intelligent enterprise demand diagnosis method and system and storage medium |
| CN115907674A (en) * | 2022-12-13 | 2023-04-04 | 广州明动软件股份有限公司 | Intelligent efficiency analysis method and system based on AI algorithm and neural engine |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102904885B (en) * | 2012-09-26 | 2015-04-29 | 北京工业大学 | Compound authentication method of multi-identity authentication information feature |
| CN108629550A (en) * | 2017-03-17 | 2018-10-09 | 武汉唐科技有限公司 | It is a kind of can backstage automatic running the office automation system implementation method |
| CN107133785A (en) * | 2017-06-12 | 2017-09-05 | 山东浪潮云服务信息科技有限公司 | A kind of pending business prompting method of office system based on domestic CPU |
| CN109635116B (en) * | 2018-12-17 | 2023-03-24 | 腾讯科技（深圳）有限公司 | Training method of text word vector model, electronic equipment and computer storage medium |
-
2019
- 2019-08-07 CN CN201910723500.0A patent/CN110489749B/en active Active
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111782811A (en) * | 2020-07-03 | 2020-10-16 | 湖南大学 | E-government affair sensitive text detection method based on convolutional neural network and support vector machine |
| CN112712177A (en) * | 2020-12-29 | 2021-04-27 | 上海永骁智能技术有限公司 | Knowledge engineering method and device based on cooperative processing |
| CN114331226A (en) * | 2022-03-08 | 2022-04-12 | 天津联创科技发展有限公司 | Intelligent enterprise demand diagnosis method and system and storage medium |
| CN115907674A (en) * | 2022-12-13 | 2023-04-04 | 广州明动软件股份有限公司 | Intelligent efficiency analysis method and system based on AI algorithm and neural engine |
| CN115907674B (en) * | 2022-12-13 | 2023-12-01 | 广州明动软件股份有限公司 | Intelligent efficiency analysis method and system based on AI algorithm and neural engine |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110489749B (en) | 2021-05-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Yang et al. | Social media data analytics for business decision making system to competitive analysis | |
| CN110489749A (en) | Intelligent Office-Automation System Work Flow Optimizing | |
| US20130151305A1 (en) | Method and Apparatus for Business Drivers and Outcomes to Enable Scenario Planning and Simulation | |
| CN109491642A (en) | A kind of Requirements Modeling system and method based on scene, information data processing terminal | |
| US20060075391A1 (en) | Distributed scenario generation | |
| JP2008544407A (en) | Technical methods and tools for capability-based multiple family of systems planning | |
| CN114385150B (en) | Low code development method based on intelligent engine | |
| CN110490547A (en) | Office system intellectualized technology | |
| US8051154B2 (en) | Enterprise service delivery technical framework | |
| CN113902394A (en) | Service platform for realizing artificial intelligence industry ecology | |
| CN115686280A (en) | Deep learning model management system, method, computer device and storage medium | |
| US20120078967A1 (en) | Integration of a Framework Application and a Task Database | |
| CN113421056A (en) | Internet human resource management system | |
| CN110489667A (en) | Intelligent circulation of official document technology based on user's portrait | |
| CN116739408A (en) | Power grid dispatching safety monitoring method and system based on data tag and electronic equipment | |
| Nielsen et al. | Level of visualization support for project communication in the Turkish construction industry: A quality function deployment approach | |
| CN114924879A (en) | Online design service system for enterprise marketing planning scheme | |
| KR20060012572A (en) | System and methods for managing distributed design chains | |
| Li et al. | Enterprise precision marketing effectiveness model based on data mining technology | |
| CN112906683A (en) | Text labeling method, device and equipment | |
| Shen et al. | Enhance the evaluation quality of project performance based on fuzzy aggregation weight effect | |
| Adikara et al. | Combining Organization Goal-Oriented Elicitation and Wireframe for Fishery Product Sales Mobile Application Design | |
| Polat | Digital Transformation Process in Selection and Implementation of an Electronic Document Management System | |
| Wang | Design and Simulation of Human Resource Allocation Model Based on Artificial Intelligence and In-Depth Data Analysis | |
| Amissah et al. | Data-centric development of architecture models with the DM2 and MS-SDF |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |