CN110009242A - Tender Evaluation Method neural network based and device, storage medium - Google Patents
Tender Evaluation Method neural network based and device, storage medium Download PDFInfo
- Publication number
- CN110009242A CN110009242A CN201910292376.7A CN201910292376A CN110009242A CN 110009242 A CN110009242 A CN 110009242A CN 201910292376 A CN201910292376 A CN 201910292376A CN 110009242 A CN110009242 A CN 110009242A
- Authority
- CN
- China
- Prior art keywords
- bid
- bidding
- data
- assessment
- project
- 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
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
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- 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
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/08—Auctions
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Human Resources & Organizations (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Strategic Management (AREA)
- General Physics & Mathematics (AREA)
- Entrepreneurship & Innovation (AREA)
- Economics (AREA)
- Development Economics (AREA)
- Marketing (AREA)
- Finance (AREA)
- Accounting & Taxation (AREA)
- Educational Administration (AREA)
- General Business, Economics & Management (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computational Linguistics (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Health & Medical Sciences (AREA)
- Game Theory and Decision Science (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Tourism & Hospitality (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention provides a kind of tender Evaluation Method neural network based and device, storage medium.This method comprises: the bid object data of acquisition project for bidding, then, the bid object data is pre-processed, so that the bid object data meets the input requirements of bid assessment models, to, bid assessment is carried out to the bid object data using the bid assessment models, with the acceptance of the bid object of the determination project for bidding.Technical solution of the present invention can effectively avoid a possibility that by malicious user black-box operation, be conducive to improve Bidding Evaluation process objective and accurate property with it is fair and just.
Description
Technical field
The present invention relates to data processing technique more particularly to a kind of tender Evaluation Method neural network based and devices, storage
Medium.
Background technique
Along with economic fast development, calls for bid as a kind of buying means to come into the open, more and more frequently appear in
In people's lives.For tenderer, this buying means are conducive to optimum selecting, improve itself competitiveness.
In traditional bidding mode, after tenderer's external disclosure project for bidding, bid material is submitted by bid direction tenderer
Material, tenderer can screen winning bidder according to the bid material of tenderer's submission.When carrying out the assessment of bids specific to tenderer,
The staff for generally requiring tenderer analyzes these bid data and assesses marking, thus, comprehensive each staff
Marking score obtain each tenderer's score, tenderer with each tenderer's score for main foundation, artificial selection winning bidder.
Existing tender Evaluation Method mainly realizes the assessment of bids in a manner of manually scoring, and the subjective factor of staff directly determines
Bidder whether may acceptance of the bid, this fair and just of assessment of bids mode be difficult to ensure, it is easy to by malicious user black-box operation.
Summary of the invention
The present invention provides a kind of tender Evaluation Method neural network based and device, storage medium, to a certain extent
It solves artificial assessment of bids bring and loses equity problem.
In a first aspect, the present invention provides a kind of tender Evaluation Method neural network based, comprising:
Acquire the bid object data of project for bidding；
The bid object data is pre-processed, so that the bid object data meets bid assessment models
Input requirements；
Bid assessment is carried out to the bid object data using the bid assessment models, with the determination project for bidding
Acceptance of the bid object.
Second aspect, the present invention provide a kind of assessment of bids device neural network based, comprising:
Acquisition module, for acquiring the bid object data of project for bidding；
Preprocessing module, for being pre-processed to the bid object data, so that the bid object data is full
The input requirements of foot bid assessment models；
Call for bid evaluation module, for carrying out bid assessment to the bid object data using the bid assessment models,
With the acceptance of the bid object of the determination project for bidding.
The third aspect, the present invention provide a kind of assessment of bids device neural network based, comprising:
Memory；
Processor；And
Computer program；
Wherein, the computer program stores in the memory, and is configured as being executed by the processor with reality
Now method as described in relation to the first aspect.
Fourth aspect, the present invention provide a kind of computer readable storage medium, are stored thereon with computer program,
The computer program is executed by processor to realize method as described in relation to the first aspect.
Tender Evaluation Method neural network based provided by the invention and device, storage medium, for project for bidding, by adopting
Collect bid object data, and it is pre-processed, thus, bid assessment is realized by trained bid assessment models,
More met the bid object of project for bidding demand using this as acceptance of the bid object.This processing mode passes through neural network mould
Type is realized, is conducive to improve a possibility that capable of effectively avoiding by malicious user black-box operation compared to the mode of the artificial assessment of bids
The objective and accurate property of Bidding Evaluation process with it is fair and just.
Detailed description of the invention
The drawings herein are incorporated into the specification and forms part of this specification, and shows the implementation for meeting the disclosure
Example, and together with specification for explaining the principles of this disclosure.
Fig. 1 is a kind of flow diagram of tender Evaluation Method neural network based provided by the embodiment of the present invention；
Fig. 2 is the flow diagram of another kind tender Evaluation Method neural network based provided by the embodiment of the present invention；
Fig. 3 is a kind of schematic network structure of Recognition with Recurrent Neural Network model provided by the embodiment of the present invention；
Fig. 4 is a kind of configuration diagram of container cloud provided by the embodiment of the present invention；
Fig. 5 is the flow diagram of another kind tender Evaluation Method neural network based provided by the embodiment of the present invention；
Fig. 6 is a kind of functional block diagram of assessment of bids device neural network based provided by the embodiment of the present invention；
Fig. 7 is a kind of entity structure schematic diagram of assessment of bids device neural network based provided by the embodiment of the present invention.
Through the above attached drawings, it has been shown that the specific embodiment of the disclosure will be hereinafter described in more detail.These attached drawings
It is not intended to limit the scope of this disclosure concept by any means with verbal description, but is by referring to specific embodiments
Art technology object person illustrates the concept of the disclosure.
Specific embodiment
Example embodiments are described in detail here, and the example is illustrated in the accompanying drawings.Following description is related to
When attached drawing, unless otherwise indicated, the same numbers in different drawings indicate the same or similar elements.Following exemplary embodiment
Described in embodiment do not represent all implementations consistent with this disclosure.On the contrary, they be only with it is such as appended
The example of the consistent device and method of some aspects be described in detail in claims, the disclosure.
Noun according to the present invention is explained first:
Bid: refer to a kind of screening and confirmation that buying side is carried out using the means to come into the open.
The assessment of bids: refer to the tender documents provided according to tenderer to screen winning bidder.
It should be noted that bid object and bid object, for a specific bid task, bid is appointed
The bid object of business A can be used as a bid object in another bid task B, and the bid object for the task A that calls for bid is another
It can be used as bid object in one bid task C.
The targeted application scenarios of the embodiment of the present invention are as follows: Bidding Evaluation scene, it is, how to determine project for bidding
The scene of winning bidder.
As previously mentioned, traditional Bidding Evaluation project is to be realized with the artificial scoring of bid person for main foundation, it is this
Links in assessment of bids mode have the winning bidder for being led to Bidding Evaluation prediction of result by the possibility of malicious user black-box operation
It is lower with the matching degree of project for bidding, damage the actual benefit of tenderer.
And technical solution provided by the invention, it is intended to solve the technical problem as above of the prior art, and propose following solve
Thinking: bid object data is handled using the bid assessment models that neural network model is foundation training, passes through automation
Processing, avoids the malicious interference of manual intervention process, is desirably to obtain the acceptance of the bid object for more meeting project for bidding demand.
How to be solved with technical solution of the specifically embodiment to technical solution of the present invention and the application below above-mentioned
Technical problem is described in detail.These specific embodiments can be combined with each other below, for the same or similar concept
Or process may repeat no more in certain embodiments.Below in conjunction with attached drawing, the embodiment of the present invention is described.
Embodiment one
The embodiment of the invention provides a kind of tender Evaluation Methods neural network based.Referring to FIG. 1, this method includes as follows
Step:
S102 acquires the bid object data of project for bidding.
Under normal circumstances, the tender documents that bid object is provided to bid object generally comprise the brief general of own situation
State and for project for bidding bidding condition.But these data are as the unsolicited data of bid object, authenticity
Need to be investigated, and the case where bid object itself is less, therefore, in actual realization scene, which throws except acquisition
Mark object provide tender documents except, can also the data very to bid object own situation be acquired.
In a kind of possible design, bid object data involved by the embodiment of the present invention be can include but is not limited to
It is following at least one: bid object be good at field, the degree of correlation for being good at field and the affiliated industry of the project for bidding,
At least one of scale data, asset data, employee's data and collage-credit data.
Wherein, bid object is good at field, the degree of correlation for being good at field Yu the affiliated industry of the project for bidding
Can mutually it agree in terms of industry field for characterizing bid object and project for bidding；And scale data, the assets of bid object
Data, employee's data can be used in characterizing the strength of enterprise of bid object；And collage-credit data is used to characterize the sincere situation of enterprise,
It may particularly include but is not limited to: whether bid object has illegal record；If there is illegal record, the type and shadow of the illegal record
Ring data etc..
When specifically executing the data collection steps, tender documents are submitted from the active of bid object, and for this
Inventive embodiments other bid object datas above-mentioned, then can efficiently grab the political affairs of each department by the Distributed Architecture of open source
Mansion bidding website, the competitive bidding information of each enterprise, open details, the disclosure of bid object of the object that calls for bid are believed in detail
Breath, on the open details of bid object or relevant other objects (company or enterprise) of bid object, reference website respectively
The open reference information of enterprise.
S104 pre-processes the bid object data, so that the bid object data meets bid assessment
The input requirements of model.
The embodiment of the present invention realizes that the bid to each bid object is assessed by trained bid assessment models, comments
Estimating foundation is collected bid object data in aforementioned S102 step.As previously mentioned, bid object data includes multiple types,
The dimension of these data is inconsistent, therefore, can be according to predetermined format pair for the ease of the training and application of evaluation of bids model
Bid object data is pre-processed, using by the data processing of literal type as digital data, to meet bid assessment models
Input requirements.
Specifically, can use preset normalization algorithm when executing the step and handle the bid object data,
To carry out vectorization processing to the bid object data after normalization, obtain the bid feature vector of the bid object.It changes
Yan Zhi, using bid feature vector as the input of bid assessment models.
Wherein, the mode of normalized can according to need customized setting.In a kind of possible design, use
" min-max standardization " mode of algorithm as normalized, aforementioned each bid object data is normalized between [0,1]
Numerical value, and the corresponding numerical value of the corresponding each bid object data of each bid object is subjected to vectorization processing, M can be obtained
The feature vector of dimension (M is the number of types for the bid object data that abovementioned steps obtain, for the integer greater than 0).
S106 carries out bid assessment to the bid object data using the bid assessment models, with the determination trick
The acceptance of the bid object of mark project.
Specifically, bid assessment is carried out to bid object data using bid assessment models, it is, the object that will submit a tender
Data input the trained bid assessment models, to obtain the bid assessment result of its output, thus, based on bid assessment knot
Fruit is the acceptance of the bid object that can determine project for bidding.
In a kind of possible design, aforementioned bid assessment models can be the bid only for project for bidding described in S102
Assessment models, as a result, before executing S106 step, which has been directed to the project for bidding and has completed training, as a result,
Bid object data (feature vector for the bid object that S104 is obtained) directly can be inputted into the bid assessment models, can be obtained
Call for bid assessment result.The bid assessment models that this design method training obtains more are bonded practical project for bidding, are conducive to improve
The accuracy rate for the assessment result that calls for bid.
In alternatively possible design, aforementioned bid assessment models can be versatility suitable for any project for bidding
Call for bid assessment models, as a result, before executing S106 step, it is also necessary to further combined with the bid number of tasks of the bid task
According in this way, trick can be obtained in will call for bid task data and bid object data, the together input as the bid assessment models
Mark assessment result.The design of this bid assessment models is more for versatility, without setting respectively to each different bid task
Meter bid assessment models, have higher flexibility, and in practical application scene, are conducive to reduced-order models training process and bring
Duration waste, improve treatment effeciency.
Wherein, project for bidding data are used to characterize the practical bidding requirement of project for bidding.In concrete implementation scene,
It can include but is not limited to following at least one data: item types, project demands information, affiliated category of employment and owned enterprise
Information.Wherein, owned enterprise's information is used to describe the company information of the bid object, may particularly include but is not limited to: scale number
According to, asset data and employee's data.
As previously mentioned, the input of bid assessment models can be designed as: only inputting bid object data；Alternatively, design are as follows:
Input bid object data and project for bidding data.
On the other hand, the output data for the assessment models that call for bid is used to indicate whether bid object gets the bid.It is specifically setting as a result,
When meter bid assessment models, output data can include but is not limited to following at least one:
The acceptance of the bid object of the project for bidding；
It is used to indicate the whether middle target first of bid object and indicates information；
It is used to indicate the second indication information of bid object tender probability.
In addition it is also necessary to explanation, bid assessment models involved by the embodiment of the present invention can be used for one or
Multiple bid objects carry out bid assessment.That is, bid assessment models can call for bid for each bid object respectively
Assessment, and obtain bid assessment result.This mode individually handled is suitable for bid object since the data volume of processing is less
Carry out the scene of bid assessment under less scene to each bid object respectively.Alternatively, bid assessment models can also be at least
Two bid objects carry out bid assessment simultaneously, and obtain bid assessment result.This design is more suitable for multiple bid objects
Scene, processing while can be realized by a bid assessment models to multiple bid objects, be conducive to improve processing effect
Rate.
In order to make it easy to understand, hereinafter, with the corresponding three bid objects of project for bidding X: bid object A, bid object B with
For bid object C, it is illustrated.
In a kind of possible design, bid assessment models are the independent processing model for project for bidding X, can be simultaneously
Multiple bid objects are handled, then it can be defeated by bid object A, bid object B and the bid object data of bid object C
Enter the assessment models that call for bid, which then can directly export acceptance of the bid object, e.g., if output bid object B, then it represents that throw
Marking object B is acceptance of the bid object.
In alternatively possible design, bid assessment models are universal model, then can be by the project for bidding of project for bidding X
Input of the bid object data of data and bid object A as the bid assessment models, the bid assessment models are exportable should
Whether bid object A is acceptance of the bid object.Such as, it can directly export: yes/no, it is positive or negative, etc..First instruction information of output
With bid object whether get the bid between corresponding relationship can customize setting.
In alternatively possible design, bid assessment models are the independent processing model for project for bidding X, can be same
When multiple bid objects are handled, then can be by the bid object data of bid object A, bid object B and bid object C
Input bid assessment models, the bid assessment models then can directly export the tender probability of each bid object.
With specific reference to the bid assessment result come determine acceptance of the bid object when, can by externally exporting the tender probability, with
It can be artificial to choose and determine winning bidder according to the tender probability of each bid object convenient for tenderer.In this way, passing through neural network
Automatic processing cooperates artificial flexible choice, can not only be met project for bidding demand, but also meet the acceptance of the bid of tenderer's regard
Person has higher freedom degree and flexibility.Alternatively, tender probability sequence can also be selected to lean on according to the sequence of tender probability
Preceding one or more bid objects, using as final acceptance of the bid object.
It is found that bid assessment models provided in an embodiment of the present invention can be appointed based on aforementioned input data, output data
Anticipate Combination Design, aforementioned each implementation be only used for for example, not to limit bid assessment models range.
And the use of aforementioned any bid assessment models, before executing S106 step, it is also necessary to pass through following steps: instruction
Practice bid assessment models.
In a kind of possible design, referring to FIG. 2, can also include into training step of playing before executing S106:
S1052 obtains bidding sample data.
Wherein, related to the bid design of assessment models, if it is only used for carrying out the independent assessment of bids for aforementioned project for bidding,
Sample bid object data is then only obtained as bidding sample data；Alternatively, if the bid assessment models be for appoint
The all applicable universal model of project for bidding of anticipating, then need to obtain sample bid object data and sample project for bidding data, to make
For bidding sample data.
S1054 utilizes the bidding sample data, training Recognition with Recurrent Neural Network model (Recurrent Neural
Network, RNN), until obtaining preset cycle-index.
It is, bid assessment models described in the embodiment of the present invention are obtained by Recognition with Recurrent Neural Network RNN model training
It arrives.
S1056 obtains the smallest Recognition with Recurrent Neural Network model of penalty values in each circulation, to assess mould as the bid
Type.
Please refer to the network structure of RNN model shown in Fig. 3.As shown in figure 3, X indicates the input of training pattern；S table
Show hidden layer；The output of O expression training pattern；U, W, V are used to indicate the weight parameter of training pattern；What t was indicated is state.
It can be seen that RNN model usually has up of three-layer, it is input layer, hidden layer and output layer respectively, there are one for the hidden layer of RNN
The oriented feedback side of item, exactly this feedback mechanism impart RNN memory capability.
Based on by RNN model when training bid assessment models, by the bidding sample data after normalized
Input X of the feature vector as the RNN training pattern, and the input S of hidden layertThere are two sources, and one is current XtIt is defeated
Enter, the other is the output S of Last status hidden layert-1, as a result, can be by representation as shown in Figure 3 are as follows:
Wherein, g indicates the activation primitive of output layer, and f is the activation primitive of hidden layer, and hidden layer is a circulation layer.
In a kind of possible design, g function can be soft max function；And f function then can be tanh function.With
Under, by taking the design as an example, illustrate the training process of RNN training pattern.
Firstly, the initiation parameter of initialization RNN training pattern.At this point, can then be incited somebody to action since f function is tanh function
The initiation parameter of RNN training pattern is initialized as the link number that n is preceding layer access, and is passed in circulation layer using time reversal
Broadcast (Backpropagation Through Time, BPTT) algorithm training parameter.Process using BPTT training parameter is as follows:
Firstly, the output valve of each hidden layer of forward calculation, then, the error entry value of each hidden layer of retrospectively calculate calculate every in turn
The gradient of a weight parameter, finally, updating weight parameter with stochastic gradient descent algorithm again.
In the embodiment of the present invention, using loss function index come the accuracy of evaluation model, loss function is pre- for measuring
The difference condition between result and legitimate reading is surveyed, that is, loss function value is smaller, the difference between prediction result and legitimate reading
Different smaller, the robustness for the bid evaluation model that training obtains is better.
In specific implementation, can be using cross entropy as loss function, if there is N number of sample, loss function can be with table
It is shown as:
Wherein, y is true value, and o is the predicted value of model.
Therefore, when executing the training of bid assessment models, default iteration can reached by way of iterative cycles
Stop iterative cycles when cycle-index, and choose corresponding parameter when wherein loss function value minimum, obtains trained bid
Assessment models.
By as above designing, technical solution provided by the embodiment of the present invention can be realized certainly using neural network model
It is dynamic to realize bid assessment, and have higher robustness and accuracy rate, this can be avoided tradition bid assessment to a certain extent
Gray zone in the process, and improve the fair and just of bid evaluation process.
In the embodiment of the present invention, after getting bid assessment result, the bid assessment result can be directly exported, it should
The acceptance of the bid object of the assessment result that calls for bid prediction is directly as acceptance of the bid object, alternatively, being only used for providing reference for bid object, with more
Good auxiliary tenderer makes better decision.
In a kind of possible design, Tensorflow can be generated into RESTFUL in conjunction with Flask frame
(Representational State Transfer FUL) interface simultaneously calls, and by front end page, carries out result displaying.Its
In, TensorFlow is the symbolic mathematical system based on data flow programming (dataflow programming), extensive
Programming applied to all kinds of machine learning (machine learning) algorithm is realized, and Flask frame is a use
The lightweight network Web application framework that Python writes, RESTFUL interface is that one kind is mainly used between client and server
Interactive class interface.
In addition, container cloud, which when being disposed, can be used, in technical solution provided by the embodiment of the present invention carries out one
Standing posture disposes O&M, and to promote the performance of intelligent invitation system, also, container cloud naturally supports the frame of elastic telescopic
Structure, and have the characteristics such as vertical dilatation, gray scale upgrading, service discovery, service orchestration and performance monitoring.
Fig. 4 and Fig. 5 are please referred to as a result, Fig. 4 shows a kind of configuration diagram of container cloud, as shown in figure 4,
In the deployment framework of container cloud, comprising a supervisor and multiple servers (Fig. 4 shows 4 servers), supervisor is used for
Image file being generated and sent to server, server receives can construct container after image file according to image file, and
The method flow indicated in image file is executed in container.
At this point, Fig. 5 shows the realization stream that server side executes the bid evaluation scheme of the aforementioned offer of the embodiment of the present invention
Journey, at this point, including the following steps:
S502 receives the image file that supervisor is sent.
S504 constructs container according to the image file, to execute any implementation as previously described in the above-described container
In at least one step.
It is, aforementioned any implementation institute providing method process can be executed in the container based on image file
In at least one step.And execute which step in a reservoir, then it can be determined by the program for including in image file,
It is, can be determined by supervisor.
For the step shown in Fig. 1, if the first program is in the above-described container, acquiring the bid number of objects of project for bidding
According to；And the second program is in the above-described container, calling for bid to the bid object data using the bid assessment models
Assessment, with the acceptance of the bid object of the determination project for bidding.
So, it in a kind of possible design, if in image file only including the first program, is constructed based on image file
Container in, execute S102 Tender Based object data acquisition step, and by obtained bid object data export, in order to hold
The subsequent S104 of row and later step.
In alternatively possible design, if only including the second program in image file, treated by aforementioned S102, S104
Data input pod, and in the container constructed based on image file, execute bid appraisal procedure described in S104.
In alternatively possible design, if being completed in a reservoir in image file comprising the first program and the second program
It is exported after S102, input pod, and the bid appraisal procedure in container in completion S104 again again after pretreatment is realized outside container.
In alternatively possible design, image file includes the program of whole processes as shown in Figure 1, then based on mirror image text
In the container of part building, whole processes shown in FIG. 1 are completed.
In addition, the training of bid assessment models as shown in Figure 2 can also execute in a reservoir, repeat no more.
In the embodiment of the present invention, container cloud can be specially Kubernetes container cloud.Wherein, Kubernetes
Also referred to as K8s is an open source, for managing the application of the containerization in cloud platform in multiple main frames, has removable
It plants, is expansible, the good characteristic of automation.
Specifically, supervisor realizes elastic telescopic framework by the management to server side.Actually utilizing container
When cloud realizes system architecture, as shown in figure 3, a supervisor can be used for managing the opening and closing of multiple server upper containers.Such as
This can be according to the actual conditions of bid evaluation process, adjustment bid assessment scale by container cloud.
Specifically, when needing to expand bid assessment scale, or when current bid assessment pressure is larger, supervisor can be sought
The lower server of present load is looked for, and image file is sent to the server, at this point, the server can be according to image file
Start a new container, and execute foregoing schemes, this one-touch can expand the upper loading limit serviced, realize to the one of cluster
Keyed dilatation, this hardly brings the increase of any O&M burden.
And when needing to reduce scale or server stress is smaller when needing recovery section resource, manager only needs to close
The container started in part server therein is closed, the recycling to this part server resource can be realized.
It is stretched framework based on aforementioned flexible, container cloud can support the dynamic tune that resource is realized according to system pressure
Degree is conducive to accelerate data-handling efficiency.
In addition, also only being needed new by supervisor when needing the arbitrary steps to aforementioned bid assessment models to be adjusted
A new image file is generated, and is sent to server, and one new appearance is opened according to new image file by server
Device can be realized, and modification maintenance cost is lower, maintain easily.
In container cloud, the distributed file system Glusterfs of open source, this storage mode is can be used in bottom storage
With powerful ability extending transversely, the training number of storage number PB (Petabyte, 1024TB) capacity can be supported by extending
According to processing thousands of customers end.
It is understood that step or operation are only example, the embodiment of the present application some or all of in above-described embodiment
The deformation of other operations or various operations can also be performed.In addition, each step can be presented not according to above-described embodiment
With sequence execute, and it is possible to do not really want to execute all operationss in above-described embodiment.
Embodiment two
Tender Evaluation Method neural network based, the embodiment of the present invention provided by one further provide based on the above embodiment
Realize the Installation practice of each step and method in above method embodiment.
The embodiment of the invention provides a kind of assessment of bids devices neural network based, referring to FIG. 6, neural network should be based on
Assessment of bids device 600, comprising:
Acquisition module 61, for acquiring the bid object data of project for bidding；
Preprocessing module 62, for being pre-processed to the bid object data, so that the bid object data
Meet the input requirements of bid assessment models；
Call for bid evaluation module 63, comments for carrying out bid to the bid object data using the bid assessment models
Estimate, with the acceptance of the bid object of the determination project for bidding.
In a kind of possible design, the assessment of bids device 600 neural network based further include: (Fig. 6 is not for model training module
Show), model training module is specifically used for:
Bid assessment is being carried out to the bid object data using the Evaluating Bidding Model, with the determination project for bidding
It gets the bid before object, obtains bidding sample data；
Utilize the bidding sample data, training Recognition with Recurrent Neural Network model, until obtaining preset cycle-index；
The smallest Recognition with Recurrent Neural Network model of penalty values in each circulation is obtained, using as the bid assessment models.
In the embodiment of the present invention, the input data of the bid assessment models, comprising:
Bid object data；Alternatively,
Bid object data and project for bidding data.
In the embodiment of the present invention, the project for bidding data include following at least one: item types, project demands letter
Breath, affiliated category of employment and owned enterprise's information.
In the embodiment of the present invention, the bid object data includes following at least one: bid object is good at field, institute
State degree of correlation, scale data, asset data, employee's data and the reference number in the field of being good at and the affiliated industry of the project for bidding
At least one of according to.
In the embodiment of the present invention, the output data of the bid assessment models, including following at least one:
The acceptance of the bid object of the project for bidding；
It is used to indicate the whether middle target first of bid object and indicates information；
It is used to indicate the second indication information of bid object tender probability.
In the embodiment of the present invention, the bid assessment models are used to carry out calling for bid to one or more bid object datas and comment
Estimate.
In a kind of possible design, preprocessing module 62 is specifically used for:
The bid object data is handled using preset normalization algorithm；
Vectorization processing is carried out to the bid object data after normalization, obtain the bid feature of the bid object to
Amount.
In a kind of possible design, the assessment of bids device 600 neural network based further include: (Fig. 6 does not show container cloud module
Out), the container cloud module, is specifically used for:
Before the bid object data of the acquisition project for bidding, the image file that supervisor is sent is received；
Container is constructed according to the image file, to execute the method as described in preceding any implementation in the above-described container
In at least one step.
Specifically, including the first program in the image file in alternatively possible design, first program is used
In: in the above-described container, acquire the bid object data of project for bidding；
And/or
It include the second program in the image file, second program is used for: in the above-described container, utilizing the bid
Assessment models carry out bid assessment to the bid object data, with the acceptance of the bid object of the determination project for bidding.
The assessment of bids device 600 neural network based of embodiment illustrated in fig. 6 can be used for executing the skill of above method embodiment
Art scheme, implementing principle and technical effect can be with further reference to the associated descriptions in embodiment of the method, and optionally, this is based on
The assessment of bids device 600 of neural network can be with server (server matched described in earlier figures 4 with supervisor).
It should be understood that the division of the modules of assessment of bids device 600 neural network based shown in figure 6 above is only one kind
The division of logic function can be completely or partially integrated on a physical entity in actual implementation, can also be physically separate.
And these modules can be realized all by way of processing element calls with software；It can also be all real in the form of hardware
It is existing；It can realize that part of module passes through formal implementation of hardware by way of processing element calls with part of module with software.
For example, bid evaluation module 63 can be the processing element individually set up, it also can integrate and filled in the assessment of bids neural network based
It sets in 600, such as is realized in some chip of terminal, in addition it is also possible to be stored in the form of program based on neural network
Assessment of bids device 600 memory in, called and executed by some processing element of assessment of bids device 600 neural network based
The function of the above modules.The realization of other modules is similar therewith.Furthermore these modules completely or partially can integrate one
It rises, can also independently realize.Processing element described here can be a kind of integrated circuit, the processing capacity with signal.?
During realization, each step of the above method or the above modules can pass through the integration logic of the hardware in processor elements
The instruction of circuit or software form is completed.
For example, the above module can be arranged to implement one or more integrated circuits of above method, such as:
One or more specific integrated circuits (Application Specific Integrated Circuit, ASIC), or, one
Or multi-microprocessor (digital singnal processor, DSP), or, one or more field programmable gate array
(Field Programmable Gate Array, FPGA) etc..For another example, when some above module dispatches journey by processing element
When the form of sequence is realized, which can be general processor, such as central processing unit (Central Processing
Unit, CPU) or it is other can be with the processor of caller.For another example, these modules can integrate together, with system on chip
The form of (system-on-a-chip, SOC) is realized.
Also, the embodiment of the invention provides a kind of assessment of bids devices neural network based, referring to FIG. 7, should be based on mind
Assessment of bids device 700 through network, comprising:
Memory 710；
Processor 720；And
Computer program；
Wherein, computer program is stored in memory 710, and is configured as being executed by processor 720 to realize as above
State method described in embodiment.
Wherein, the number of processor 720 can be one or more, processing in assessment of bids device 700 neural network based
Device 720 is referred to as processing unit, and certain control function may be implemented.The processor 720 can be general processor
Or application specific processor etc..In a kind of optionally design, processor 720 can also have instruction, and described instruction can be by institute
The operation of processor 720 is stated, so that the assessment of bids device 700 neural network based executes side described in above method embodiment
Method.
In another possible design, assessment of bids device 700 neural network based may include circuit, and the circuit can
To realize the function of sending or receiving or communicate in preceding method embodiment.
Optionally, the number of memory 710 can be one or more in the assessment of bids device 700 neural network based
It is a, there are instruction or intermediate data on memory 710, described instruction can be run on the processor 720, so that described
Assessment of bids device 700 neural network based executes method described in above method embodiment.Optionally, the memory 710
In can also be stored with other related datas.Optionally it also can store instruction and/or data in processor 720.The processing
Device 720 and memory 710 can be separately provided, and also can integrate together.
In addition, as shown in fig. 7, being additionally provided with transceiver 730 in the assessment of bids device 700 neural network based, wherein
The transceiver 730 is properly termed as Transmit-Receive Unit, transceiver, transmission circuit or transceiver etc., for test equipment or its
His terminal device carries out data transmission or communicates, and details are not described herein.
As shown in fig. 7, memory 710, processor 720 are connected and communicated with transceiver 730 by bus.
If the assessment of bids device 700 neural network based is for realizing the method corresponded in Fig. 2, for example, can be by
The image file that transceiver 730 is sent to reception supervisor.And processor 720 determines accordingly or controls behaviour for completing
Make, optionally, corresponding instruction can also be stored in memory 710.The specific processing mode of all parts can refer to
The associated description of previous embodiment.
In addition, it is stored thereon with computer program the embodiment of the invention provides a kind of readable storage medium storing program for executing, the computer
Program is executed by processor to realize the method as described in embodiment one.
The part that the present embodiment is not described in detail can refer to the related description to embodiment one.
Those skilled in the art after considering the specification and implementing the invention disclosed here, will readily occur to its of the disclosure
Its embodiment.The present invention is directed to cover any variations, uses, or adaptations of the disclosure, these modifications, purposes or
Person's adaptive change follows the general principles of this disclosure and including the undocumented common knowledge in the art of the disclosure
Or conventional techniques.The description and examples are only to be considered as illustrative, and the true scope and spirit of the disclosure are by following
Claims are pointed out.
Claims (13)
1. a kind of tender Evaluation Method neural network based characterized by comprising
Acquire the bid object data of project for bidding；
The bid object data is pre-processed, so that the bid object data meets the input of bid assessment models
It is required that；
Bid assessment is carried out to the bid object data using the bid assessment models, in the determination project for bidding
Mark object.
2. the method according to claim 1, wherein described utilize the Evaluating Bidding Model to the bid number of objects
According to carrying out bid assessment, before the acceptance of the bid object of the determination project for bidding, the method also includes:
Obtain bidding sample data；
Utilize the bidding sample data, training Recognition with Recurrent Neural Network model, until obtaining preset cycle-index；
The smallest Recognition with Recurrent Neural Network model of penalty values in each circulation is obtained, using as the bid assessment models.
3. method according to claim 1 or 2, which is characterized in that the input data of the bid assessment models, comprising:
Bid object data；Alternatively,
Bid object data and project for bidding data.
4. according to the method described in claim 3, it is characterized in that, the project for bidding data include following at least one: item
Mesh type, project demands information, affiliated category of employment and owned enterprise's information.
5. method according to claim 1 or 2, which is characterized in that the bid object data includes following at least one:
Bid object is good at field, degree of correlation, scale data, the assets for being good at field Yu the affiliated industry of the project for bidding
At least one of data, employee's data and collage-credit data.
6. method according to claim 1 or 2, which is characterized in that the output data of the bid assessment models, including such as
Lower at least one:
The acceptance of the bid object of the project for bidding；
It is used to indicate the whether middle target first of bid object and indicates information；
It is used to indicate the second indication information of bid object tender probability.
7. method according to claim 1 or 2, which is characterized in that the bid assessment models are used for one or more
Bid object data carries out bid assessment.
8. method according to claim 1 or 2, which is characterized in that it is described that the bid object data is pre-processed,
Include:
The bid object data is handled using preset normalization algorithm；
Vectorization processing is carried out to the bid object data after normalization, obtains the bid feature vector of the bid object.
9. the method according to claim 1, wherein it is described acquisition project for bidding bid object data before,
The method also includes:
Receive the image file that supervisor is sent；
Container is constructed according to the image file, to execute the method according to claim 1 in the above-described container
In at least one step.
10. according to the method described in claim 9, it is characterized in that,
It include the first program in the image file, first program is used for: in the above-described container, acquiring the throwing of project for bidding
Mark object data；
And/or
Include the second program in the image file, second program is used for: in the above-described container, being assessed using the bid
Model carries out bid assessment to the bid object data, with the acceptance of the bid object of the determination project for bidding.
11. a kind of assessment of bids device neural network based characterized by comprising
Acquisition module, for acquiring the bid object data of project for bidding；
Preprocessing module is recruited for pre-processing to the bid object data so that the bid object data meets
Mark the input requirements of assessment models；
Call for bid evaluation module, for carrying out bid assessment to the bid object data using the bid assessment models, with true
The acceptance of the bid object of the fixed project for bidding.
12. a kind of assessment of bids device neural network based characterized by comprising
Memory；
Processor；And
Computer program；
Wherein, the computer program stores in the memory, and is configured as being executed by the processor to realize such as
The described in any item methods of claim 1-10.
13. a kind of computer readable storage medium, which is characterized in that it is stored thereon with computer program,
The computer program is executed by processor to realize such as the described in any item methods of claim 1-10.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910292376.7A CN110009242A (en) | 2019-04-12 | 2019-04-12 | Tender Evaluation Method neural network based and device, storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910292376.7A CN110009242A (en) | 2019-04-12 | 2019-04-12 | Tender Evaluation Method neural network based and device, storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110009242A true CN110009242A (en) | 2019-07-12 |
Family
ID=67171332
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910292376.7A Pending CN110009242A (en) | 2019-04-12 | 2019-04-12 | Tender Evaluation Method neural network based and device, storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110009242A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110517123A (en) * | 2019-08-23 | 2019-11-29 | 宁夏神州好易信息发展股份有限公司 | A kind of unmanned bid evaluation system of AI based on technology of Internet of things and method |
| CN112434883A (en) * | 2020-12-12 | 2021-03-02 | 广东电力信息科技有限公司 | Expert extraction demand prediction method and device based on deep learning |
| CN112990845A (en) * | 2021-01-04 | 2021-06-18 | 江苏省测绘地理信息局信息中心 | Intelligent acquisition method for mapping market project |
| CN113449459A (en) * | 2021-04-09 | 2021-09-28 | 江西高创保安服务技术有限公司 | Universal distributed computing system design method for improving neural network reasoning accuracy and maintaining operation speed |
| CN114066336A (en) * | 2022-01-18 | 2022-02-18 | 珠海大横琴科技发展有限公司 | Bidding risk control method and device |
| CN116452306A (en) * | 2023-03-03 | 2023-07-18 | 中国人民解放军军事科学院系统工程研究院 | Bid distribution method for intelligent combat task |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100153161A1 (en) * | 2004-04-27 | 2010-06-17 | Kiran Ali S | System and method for workforce requirements management |
| CN106504015A (en) * | 2016-10-17 | 2017-03-15 | 鞍钢集团矿业有限公司 | A kind of field supplier of enterprise of combination BP neural network recommends method |
| CN108428061A (en) * | 2018-03-30 | 2018-08-21 | 合肥工业大学 | Intelligent Tenders Decision system based on DEA-GA-BP and tender Evaluation Method |
| CN108764743A (en) * | 2018-06-04 | 2018-11-06 | 筑客网络技术（上海）有限公司 | A kind of automatic tender Evaluation Method for building materials buying |
| CN109255691A (en) * | 2018-08-31 | 2019-01-22 | 万翼科技有限公司 | Acquisition methods, device and the computer readable storage medium of acceptance of the bid supplier |
| CN109409693A (en) * | 2018-09-30 | 2019-03-01 | 深圳市元征科技股份有限公司 | A kind of business associate mode recommended method and relevant device |
-
2019
- 2019-04-12 CN CN201910292376.7A patent/CN110009242A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100153161A1 (en) * | 2004-04-27 | 2010-06-17 | Kiran Ali S | System and method for workforce requirements management |
| CN106504015A (en) * | 2016-10-17 | 2017-03-15 | 鞍钢集团矿业有限公司 | A kind of field supplier of enterprise of combination BP neural network recommends method |
| CN108428061A (en) * | 2018-03-30 | 2018-08-21 | 合肥工业大学 | Intelligent Tenders Decision system based on DEA-GA-BP and tender Evaluation Method |
| CN108764743A (en) * | 2018-06-04 | 2018-11-06 | 筑客网络技术（上海）有限公司 | A kind of automatic tender Evaluation Method for building materials buying |
| CN109255691A (en) * | 2018-08-31 | 2019-01-22 | 万翼科技有限公司 | Acquisition methods, device and the computer readable storage medium of acceptance of the bid supplier |
| CN109409693A (en) * | 2018-09-30 | 2019-03-01 | 深圳市元征科技股份有限公司 | A kind of business associate mode recommended method and relevant device |
Non-Patent Citations (4)
| Title |
|---|
| YOYOFU007: "RNN循环神经网络以及LSTM长短期记忆模型-简介", 《CSDN》 * |
| 冯斐: "基于深度学习的推荐算法研究", 《中国优秀硕士学位论文全文数据库 哲学与人文科学辑》 * |
| 孙修东: "基于人工神经网络的多指标综合评价方法研究", 《轻工业学院学报(自然科学版)》 * |
| 孟天宏: "基于词嵌入的个性化推荐系统的研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110517123A (en) * | 2019-08-23 | 2019-11-29 | 宁夏神州好易信息发展股份有限公司 | A kind of unmanned bid evaluation system of AI based on technology of Internet of things and method |
| CN112434883A (en) * | 2020-12-12 | 2021-03-02 | 广东电力信息科技有限公司 | Expert extraction demand prediction method and device based on deep learning |
| CN112990845A (en) * | 2021-01-04 | 2021-06-18 | 江苏省测绘地理信息局信息中心 | Intelligent acquisition method for mapping market project |
| CN113449459A (en) * | 2021-04-09 | 2021-09-28 | 江西高创保安服务技术有限公司 | Universal distributed computing system design method for improving neural network reasoning accuracy and maintaining operation speed |
| CN114066336A (en) * | 2022-01-18 | 2022-02-18 | 珠海大横琴科技发展有限公司 | Bidding risk control method and device |
| CN116452306A (en) * | 2023-03-03 | 2023-07-18 | 中国人民解放军军事科学院系统工程研究院 | Bid distribution method for intelligent combat task |
| CN116452306B (en) * | 2023-03-03 | 2023-10-10 | 中国人民解放军军事科学院系统工程研究院 | Bid distribution method for intelligent combat task |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110009242A (en) | Tender Evaluation Method neural network based and device, storage medium | |
| US10802992B2 (en) | Combining CPU and special accelerator for implementing an artificial neural network | |
| CN109240929A (en) | Software quality prediction method, apparatus, terminal and computer readable storage medium | |
| JP2021501421A5 (en) | ||
| CN108090516A (en) | Automatically generate the method and system of the feature of machine learning sample | |
| CN108475214A (en) | adaptive artificial neural network selection technique | |
| JP6614466B2 (en) | Capability grant data generator | |
| CN110009614A (en) | Method and apparatus for output information | |
| CN107273979A (en) | The method and system of machine learning prediction are performed based on service class | |
| CN107038064A (en) | Virtual machine management method and device, storage medium | |
| CN110709862B (en) | Calculation method determination system, calculation method determination method, and recording medium | |
| CN110088775A (en) | Reinforce learning system | |
| CN113240079A (en) | Model training method and device | |
| CN109961102A (en) | Image processing method, device, electronic equipment and storage medium | |
| Burstein et al. | Artificial neural networks in decision support systems | |
| Chen et al. | Fuzzy-neural approaches with example post-classification for estimating job cycle time in a wafer fab | |
| KR102552856B1 (en) | Method, device and system for automating creation of content template and extracting keyword for platform service that provide content related to commerce | |
| Ferguson et al. | A standardized representation of convolutional neural networks for reliable deployment of machine learning models in the manufacturing industry | |
| US11762948B2 (en) | In-process engine implementation using SDK extensions | |
| US20230196204A1 (en) | Agnostic machine learning inference | |
| Nasiri et al. | The stage shop scheduling problem: lower bound and metaheuristic | |
| CN108604313A (en) | The predictive modeling of automation and frame | |
| US20230353516A1 (en) | User interaction artificial intelligence chat engine for integration of automated machine generated responses | |
| JP2020035000A (en) | Machine learning system and calculation method of boltzmann machine | |
| CN110377769A (en) | Modeling Platform system, method, server and medium based on graph data structure |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190712 |
|
| RJ01 | Rejection of invention patent application after publication |