CN109933703A - A kind of construction method of Intellectual Property Right of Enterprises appraisal Model - Google Patents
A kind of construction method of Intellectual Property Right of Enterprises appraisal Model Download PDFInfo
- Publication number
- CN109933703A CN109933703A CN201910194965.1A CN201910194965A CN109933703A CN 109933703 A CN109933703 A CN 109933703A CN 201910194965 A CN201910194965 A CN 201910194965A CN 109933703 A CN109933703 A CN 109933703A
- Authority
- CN
- China
- Prior art keywords
- data
- node
- information
- intellectual property
- enterprise
- 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.)
- Withdrawn
Links
Abstract
The invention discloses a kind of construction methods of Intellectual Property Right of Enterprises appraisal Model, belong to intellectual property value evaluation areas, construct data-acquisition system, and data-acquisition system includes that company information data obtains subsystem and Intellectual Property Right of Enterprises information acquisition subsystem.Multi-source data, which is carried out, using data-acquisition system obtains company information and intellectual property information.The company information and Intellectual Property Right of Enterprises information of vectorization enterprise respectively；Enterprise's vector characteristics tree is constructed, circular recursion neural network model is constructed.Data set is constructed, trained and assessments recurrent neural network becomes final assessment network.Use data-acquisition system as data input port, assesses network as arithmetic center and form assessment models.The present invention is a kind of succinct, accurately and efficiently vectorization method, i.e. vector characteristics tree, can fast and accurately carry out the value for assessing the intellectual property of enterprise.
Description
Technical field
The present invention relates to intellectual property value evaluation areas more particularly to a kind of Intellectual Property Right of Enterprises appraisal Models
Construction method.
Background technique
With the development of information technology, the cooperation between enterprise in terms of intellectual property is more and more closer, and exchange of skills is also got over
The information exchange for carrying out more frequent, traditional manual mode does not adapt to current enterprise need of work already, and each enterprise wishes to pass through
The means of exchange of networking realizes exchange of skills and Techno-sharing, work of this conveniently exchanged form to enterprise is improved
Efficiency has a very important significance.But the intellectual property information exchange between each enterprise is extremely complex, is not Liang Ge enterprise
Between information arbitrarily send and receive, but after transmission device and necessary security control check, and by appropriate
Format transformation is transmitted to recipient from sender.Due to the diversity of the information content, the complexity of data processing, information security
The many factors such as the timeliness that confidentiality and exchange require, so that there are certain puzzlements for the information trading between enterprise, although
Intellectual property transaction system is continuous perfect, but existing intellectual property transaction system still remains integrity service function and owes
Simply there are the loopholes such as security risk for unreasonable, technology business for scarce, system architecture, so that the use of intellectual property transaction system has
There is certain limitation.
Universal with intellectual property, many enterprises have begun attention intellectual property, while possessing to have had and certain know
Know property right, but existing intellectual property, which has, could not much carry out actual application, so that the case where the wasting of resources, therefore
Need to design a kind of method or model quickly can quickly be assessed in the valuation to the intellectual property of enterprise, to make
Enterprise's real-time tracking knows the valuation information of oneself enterprise, provide a quick and objective data to invest and finance or providing a loan etc.
It supports.
Summary of the invention
It is existing to solve the purpose of the present invention is to provide a kind of construction method of Intellectual Property Right of Enterprises appraisal Model
The technical issues of quick valuation can not reached to the intellectual property of enterprise in technology.
A kind of construction method of Intellectual Property Right of Enterprises appraisal Model, the construction method include the following steps:
Step 1: building data-acquisition system, data-acquisition system include that company information data acquisition subsystem and enterprise are known
Know property information and obtains subsystem；
Step 2: carrying out multi-source data using data-acquisition system and obtain company information and intellectual property information；
Step 3: the company information and Intellectual Property Right of Enterprises information of vectorization enterprise respectively；
Step 4: building enterprise's vector characteristics tree；
Step 5: building circular recursion neural network model；
Step 6: building data set, trained and assessments recurrent neural network become final assessment network；
Step 7: using data-acquisition system as data input port, assess network as arithmetic center and form assessment mould
Type.
Further, company information data obtains subsystem and connect with industrial and commercial bureau enterprise information system in the step 1, enterprise
Industry intellectual property information acquisition subsystem is connect with State Intellectual Property Office's patent information system, and the company information data obtains
Subsystem and Intellectual Property Right of Enterprises information acquisition subsystem include server configuration and data extractor, the server configuration
Include:
Parameter setting, provides the setting of the parameter configuration of data extractor node by a configuration server, and
When data extractor sends parameter-configuring request to server, which is returned to according to the unique number of data extractor
Configuration information；
Point extension, when increasing data extractor node, configuration server receives the node and is sent out by socket agreement
The order for the addition data extractor node brought, configuration server add a record into configuration information data, and will
The point of operation extends: when increasing data extractor node, configuration server receives the node and is sent by socket agreement
The order of the addition data extractor node to come over, configuration server add a record into configuration information data, and will fortune
The sum of capable data extractor node adds 1；
Abnormal monitoring: each data extractor node, which is sent every the 1h time to configuration server, indicates this data extractor
The order that node is operating normally, configuration server record each data extractor in data extractor node state list
The ID of node and the corresponding node that finally receives issue the time for indicating the order operated normally；When configuration server 1h
Between can the node state list of ergodic data grabber, if some data extractor node finally issues indicate operating normally
Order to current time time interval be greater than 15 minutes, then it represents that has there is exception, configuration server in the data extractor
The sum of the sum of the grabber of the operation node that subtracts 1 is added 1；
Abnormal monitoring: each data extractor node every 1h time sends to configuration server indicates this data extractor section
The order that point is operating normally, configuration server record each data extractor section in data extractor node state list
The ID of point and the corresponding node that finally receives issue the time for indicating the order operated normally；Configuration server is every 1h
Time meeting ergodic data grabber node state list, if some grabber node finally issues what expression was operating normally
The time interval of current time is ordered to be greater than 15 minutes, then it represents that exception has occurred in the data extractor, and configuration server will
The sum of the grabber of operation subtracts 1；
Load balancing: the task that each data extractor is completed is identical, and the load on data extractor refers to configuration clothes
Business device be assigned on each data extractor to grab number of videos number；Load balancing is related to two stages；First
Stage is the distribution of data, and second stage is grabbed after primary data grabber is completed by the data extractor recorded
Start and end time, calculate all grabbers and complete the times required for primary grab, new need to grab when next time has
Video when being added in list, it is according to ratio of the crawl under each grabber last registration the time required to primary that these are new
Data be assigned to each grabber node；
Data extractor includes link grabber, web-page parser, data storage and updates controller, link crawl
Device is scanned industrial and commercial bureau's enterprise information system according to breadth traversal algorithm, obtains the link of the webpage for the condition that meets, and will
These link storages are in a linked database；Meanwhile the mark whether being accessed for link addition each in database, if
Some link is by one of node visit mistake, then other then skip the node, obtains next link, if this link is not
It is accessed, then it accesses the page and the link for belonging to this website for including in the webpage is added in linked database；It is described
The webpage met the requirements refer to containing enterprise name, social Unicode, address and business scope webpage；
Web-page parser, obtains the link of all satisfactions from data storage, and load links corresponding webpage, then into
The parsing of row webpage obtains required company information；Analytic method library is constructed, if the method that parsing is not indicated in link,
It traverses each method and parses each link, when link is successfully parsed, the analytic method of the link is labeled as this method,
If having designated when being linked at using this method parsing failure of analytic method, new solution is added into analytic method library
Analysis method；
Data storage, for storing company information lists of links, enterprise's essential information and statistical data, analytic method library
Data；
The update controller, according to the update of every link in the company information lists of links stored in database frequency
Rate uses the frequency of timed task control web-page parser analyzing web page.
Further, the detailed process of the step 2 are as follows:
In same administrative division, company information data acquisition subsystem connect acquisition enterprise with industrial and commercial bureau's enterprise information system
Industry essential information, Intellectual Property Right of Enterprises information acquisition subsystem are corresponding to the connection acquisition of State Intellectual Property Office's patent information system
The intellectual property information of enterprise, intellectual property information include patent, trade mark, copyright, integrated circuit and new variety of plant；
It will be in company information and enterprise by the mark of society, affiliated enterprise Unicode for multi-source data information
Intellectual property information polymerize as unit of enterprise.
Further, the detailed process in the step 3 are as follows:
Intellectual property evaluation expertise and Feature Engineering method are introduced, data cleansing rule is established, business data is believed
Breath is screened and is identified, is rejected garbage, is merged redundancy, mark important information；
The company information and Intellectual Property Right of Enterprises information for distinguishing enterprise establish vectorization rule, realize vectorization, specific to wrap
It includes and completes dirty data processing, missing values are filled up and the normalized work of data, wherein the middle vectorization result of any object is used
A () is indicated.
Further, the detailed process in the step 4 are as follows:
By the company information of enterprise and Intellectual Property Right of Enterprises information vector as a result, root section as enterprise vector characteristics tree
Point, wherein root node is expressed as r, and poor household's vector characteristics tree table is shown as Tr, the vectorization result expression of root node are as follows:
ra=a (r)；
In enterprise vector characteristics tree TrIn, single intermediate node, i.e. vector characteristics tree T are constructed according to intellectual property classificationr
In any intermediate node p vectorization indicate paBy the y that has the rightaWith failure saTwo parts composition produces application for the knowledge of authorization
Power, with the filling node q for not including any informationaInstead of, then, point pa、ya、sa、qaIt indicates are as follows:
pa=[ya:qa]
ya=a (y ')
qa=a (q ')
sa=[0,0 ... ..0,0]
According to the actual production relationship between Intellectual Property Right of Enterprises in reality production, structure enterprise vector characteristics tree.
Further, the detailed process in the step 5 are as follows:
Deep learning model is circular recursion neural network model M, and using the recursive nature of tree, the vector of tree is indicated by it
Root node and using its child node as the vector of the subtree collection of root indicate by nonlinear operation generate, wherein vector characteristics tree
The vectorization of middle any node indicates, is provided by step 4, and the vector of subtree collection is indicated by the way that subtree is sequentially input shot and long term
Remember layer (LSTM) and calculate generation, formally, it is assumed that enterprise vector characteristics tree Tr=(V, E), wherein V and E respectively represents tree
Node set and line set.Further, it is p to any node in V, remembers that the child node collection of p is combined into C={ c1, c2...,
ci..., c|c|, and corresponding subtree collection isWherein ciForRoot node,
Then calculation formula is indicated by the vectorization of the subtree of root of p are as follows:
Encode (T)=σ (Wparent*p_encoding+Wsubtrees*Encode(Fr))
Wherein, σ () indicates activation primitive, Wparent、WsubtreesIt is parameter, p-encoding is the vectorization knot of node p
Fruit, Encode (F) are that the vectorization of subtree indicates to sequentially input LSTM layers of final output, are indicated are as follows:
It particularly points out, since the root node r of vector characteristics tree indicates user data information, in specific calculate, recurrence
The calculation formula of the terminal of operation are as follows:
Encode(Tr)=σ (Wroot*r_encoding+Wsubtrees*Encode(Tr))
Wherein, WrootIt is the known parameters in addition set, r_encoding is the vectorization of root node r as a result, Encode
(Tr) be r subtree collection vectorization as a result,
By Encode (Tr) input circular recursion neural network model in subsequent full articulamentum determined.
Further, the detailed process in the step 6 are as follows:
Data set D is constructed based on company information vector characteristics tree, wherein evaluation quantity is estimating for the existing intellectual property of enterprise
Value, evaluation quantity are enterprise's vector characteristics tree, the valuation without no longer valid intellectual property, and divide training set and test set；
Using training set D training circular recursion neural network M, and in test central evaluation M effect；
Circular recursion neural network M after training is put into real enterprise intellectual property value evaluation work and is used.
Further, the detailed process in the step 7 are as follows:
Use data-acquisition system as data input port, assesses network as arithmetic center and form assessment models, in number
Belong to enterprise's identification information according to acquisition system, enterprise name or social Unicode, data-acquisition system obtain the base of enterprise
The A to Z of property information of this information data and enterprise, and be put into assessment network and carry out generation assessment result.
Present invention employs above-mentioned technical proposal, the present invention is had following technical effect that
The present invention is a kind of succinct, accurately and efficiently vectorization method, i.e. vector characteristics tree, can fast and accurately into
The value of the intellectual property of row assessment enterprise, preferably quickly can be provided data for Corporate finance and assets assessment etc.
Support, while from various dimensions memory estimate so that estimation result it is more accurate, from multi-source data as unit of enterprise it is whole
Close data, essential information to enterprise and intellectual property information carry out vectorization, construct company information vector characteristics tree, design with
Realize circular recursion neural network model especially circular recursion neural net layer, trained and assessments recurrent neural network mould
Type simultaneously disposes use, has good practical application effect.
Detailed description of the invention
Fig. 1 is flow chart of the present invention.
Fig. 2 is circular recursion neural network model calculation schematic diagram of the present invention.
Specific embodiment
To make the objectives, technical solutions, and advantages of the present invention more comprehensible, referring to the drawings and preferred reality is enumerated
Example is applied, the present invention is described in more detail.However, it is necessary to illustrate, many details listed in specification are only to be
Reader is set to have a thorough explanation to one or more aspects of the present invention, it can also be with even without these specific details
Realize the aspects of the invention.
Referring to Fig. 1, the present invention provides a kind of construction method of Intellectual Property Right of Enterprises appraisal Model, the building side
Method includes the following steps:
Step 1: building data-acquisition system, data-acquisition system include that company information data acquisition subsystem and enterprise are known
Know property information and obtains subsystem.
Company information data obtains subsystem and connect with industrial and commercial bureau enterprise information system in the step 1, Company Knowledge production
Power information acquisition subsystem connect with State Intellectual Property Office patent information system, the company information data acquisition subsystem with
Intellectual Property Right of Enterprises information acquisition subsystem includes server configuration and data extractor, and the server configuration includes:
Parameter setting, provides the setting of the parameter configuration of data extractor node by a configuration server, and
When data extractor sends parameter-configuring request to server, which is returned to according to the unique number of data extractor
Configuration information；
Point extension, when increasing data extractor node, configuration server receives the node and is sent out by socket agreement
The order for the addition data extractor node brought, configuration server add a record into configuration information data, and will
The point of operation extends: when increasing data extractor node, configuration server receives the node and is sent by socket agreement
The order of the addition data extractor node to come over, configuration server add a record into configuration information data, and will fortune
The sum of capable data extractor node adds 1；
Abnormal monitoring: each data extractor node, which is sent every the 1h time to configuration server, indicates this data extractor
The order that node is operating normally, configuration server record each data extractor in data extractor node state list
The ID of node and the corresponding node that finally receives issue the time for indicating the order operated normally；When configuration server 1h
Between can the node state list of ergodic data grabber, if some data extractor node finally issues indicate operating normally
Order to current time time interval be greater than 15 minutes, then it represents that has there is exception, configuration server in the data extractor
The sum of the sum of the grabber of the operation node that subtracts 1 is added 1；
Abnormal monitoring: each data extractor node every 1h time sends to configuration server indicates this data extractor section
The order that point is operating normally, configuration server record each data extractor section in data extractor node state list
The ID of point and the corresponding node that finally receives issue the time for indicating the order operated normally；Configuration server is every 1h
Time meeting ergodic data grabber node state list, if some grabber node finally issues what expression was operating normally
The time interval of current time is ordered to be greater than 15 minutes, then it represents that exception has occurred in the data extractor, and configuration server will
The sum of the grabber of operation subtracts 1；
Load balancing: the task that each data extractor is completed is identical, and the load on data extractor refers to configuration clothes
Business device be assigned on each data extractor to grab number of videos number；Load balancing is related to two stages；First
Stage is the distribution of data, and second stage is grabbed after primary data grabber is completed by the data extractor recorded
Start and end time, calculate all grabbers and complete the times required for primary grab, new need to grab when next time has
Video when being added in list, it is according to ratio of the crawl under each grabber last registration the time required to primary that these are new
Data be assigned to each grabber node；
Data extractor includes link grabber, web-page parser, data storage and updates controller, link crawl
Device is scanned industrial and commercial bureau's enterprise information system according to breadth traversal algorithm, obtains the link of the webpage for the condition that meets, and will
These link storages are in a linked database；Meanwhile the mark whether being accessed for link addition each in database, if
Some link is by one of node visit mistake, then other then skip the node, obtains next link, if this link is not
It is accessed, then it accesses the page and the link for belonging to this website for including in the webpage is added in linked database；It is described
The webpage met the requirements refer to containing enterprise name, social Unicode, address and business scope webpage；
Web-page parser, obtains the link of all satisfactions from data storage, and load links corresponding webpage, then into
The parsing of row webpage obtains required company information；Analytic method library is constructed, if the method that parsing is not indicated in link,
It traverses each method and parses each link, when link is successfully parsed, the analytic method of the link is labeled as this method,
If having designated when being linked at using this method parsing failure of analytic method, new solution is added into analytic method library
Analysis method；
Data storage, for storing company information lists of links, enterprise's essential information and statistical data, analytic method library
Data；
The update controller, according to the update of every link in the company information lists of links stored in database frequency
Rate uses the frequency of timed task control web-page parser analyzing web page.
Step 2: carrying out multi-source data using data-acquisition system and obtain company information and intellectual property information.
In same administrative division, company information data acquisition subsystem connect acquisition enterprise with industrial and commercial bureau's enterprise information system
Industry essential information, Intellectual Property Right of Enterprises information acquisition subsystem are corresponding to the connection acquisition of State Intellectual Property Office's patent information system
The intellectual property information of enterprise, intellectual property information include patent, trade mark, copyright, integrated circuit and new variety of plant；
It will be in company information and enterprise by the mark of society, affiliated enterprise Unicode for multi-source data information
Intellectual property information polymerize as unit of enterprise.
Step 3: the company information and Intellectual Property Right of Enterprises information of vectorization enterprise respectively.
Intellectual property evaluation expertise and Feature Engineering method are introduced, data cleansing rule is established, business data is believed
Breath is screened and is identified, is rejected garbage, is merged redundancy, mark important information；
The company information and Intellectual Property Right of Enterprises information for distinguishing enterprise establish vectorization rule, realize vectorization, specific to wrap
It includes and completes dirty data processing, missing values are filled up and the normalized work of data, wherein the middle vectorization result of any object is used
A () is indicated.
Step 4: building enterprise's vector characteristics tree.By the company information of enterprise and Intellectual Property Right of Enterprises information vector knot
Fruit, as the root node of enterprise vector characteristics tree, wherein root node is expressed as r, and poor household's vector characteristics tree table is shown as Tr, root section
The vectorization result of point indicates are as follows:
ra=a (r)；
In enterprise vector characteristics tree TrIn, single intermediate node, i.e. vector characteristics tree T are constructed according to intellectual property classificationr
In any intermediate node p vectorization indicate paBy the y that has the rightaWith failure saTwo parts composition produces application for the knowledge of authorization
Power, with the filling node q for not including any informationaInstead of, then, point pa、ya、sa、qaIt indicates are as follows:
pa=[ya: qa]
ya=a (y ')
qa=a (q ')
sa=[0,0 ... ..0,0]
According to the actual production relationship between Intellectual Property Right of Enterprises in reality production, structure enterprise vector characteristics tree.
Step 5: building circular recursion neural network model.Deep learning model is circular recursion neural network model M, benefit
With the recursive nature of tree, the vector of tree indicates to indicate to pass through by its root node and by the vector of the subtree collection of root of its child node
Nonlinear operation generate, wherein in vector characteristics tree any node vectorization expression, provided by step 4, subtree collection to
Amount indicates to generate by the way that subtree is sequentially input shot and long term memory layer (LSTM) calculating, formally, it is assumed that enterprise's vector characteristics
Set Tr=(V, E), wherein V and E respectively represents the node set and line set of tree.Further, it is p to any node in V, remembers p
Child node collection be combined into C={ c1, c2..., ci..., c|c|, and corresponding subtree collection isWherein ciForRoot node, then indicate to count by the vectorization of the subtree of root of p
Calculate formula are as follows:
Encode (T)=σ (Wparent*p_encoding+Wsubtrees*Encode(Fr))
Wherein, σ () indicates activation primitive, Wparent、WsubtreesIt is parameter, p_encoding is the vectorization knot of node p
Fruit, Encode (F) are that the vectorization of subtree indicates to sequentially input LSTM layers of final output, are indicated are as follows:
It particularly points out, since the root node r of vector characteristics tree indicates user data information, in specific calculate, recurrence
The calculation formula of the terminal of operation are as follows:
Encode(Tr)=σ (Wroot*r_encoding+Wsubtrees*Encode(Tr))
Wherein, WrootIt is the known parameters in addition set, r_encoding is the vectorization of root node r as a result, Encode
(Tr) be r subtree collection vectorization as a result,
By Encode (Tr) input circular recursion neural network model in subsequent full articulamentum determined.
Step 6: building data set, trained and assessments recurrent neural network become final assessment network.Based on enterprise
Information vector characteristics tree construct data set D, wherein evaluation quantity be the existing intellectual property of enterprise valuation, evaluation quantity be enterprise to
Measure feature tree, the valuation without no longer valid intellectual property, and divide training set and test set；
Using training set D training circular recursion neural network M, and in test central evaluation M effect；
Circular recursion neural network M after training is put into real enterprise intellectual property value evaluation work and is used.
Step 7: using data-acquisition system as data input port, assess network as arithmetic center and form assessment mould
Type.
Use data-acquisition system as data input port, assesses network as arithmetic center and form assessment models, in number
Belong to enterprise's identification information according to acquisition system, enterprise name or social Unicode, data-acquisition system obtain the base of enterprise
The A to Z of property information of this information data and enterprise, and be put into assessment network and carry out generation assessment result.
The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention, for this field skill
For art personnel, it is clear that invention is not limited to the details of the above exemplary embodiments, and without departing substantially from spirit of the invention or
In the case where essential characteristic, the present invention can be realized in other specific forms.Therefore, in all respects, should all incite somebody to action
Embodiment regards exemplary as, and is non-limiting, the scope of the present invention by appended claims rather than on state
Bright restriction, it is intended that including all changes that fall within the meaning and scope of the equivalent elements of the claims in the present invention
It is interior.Any reference signs in the claims should not be construed as limiting the involved claims.
The above is only a preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art
For member, without departing from the principle of the present invention, it can also make several improvements and retouch, these improvements and modifications are also answered
It is considered as protection scope of the present invention.
Claims (8)
1. a kind of construction method of Intellectual Property Right of Enterprises appraisal Model, which is characterized in that the construction method includes as follows
Step:
Step 1: building data-acquisition system, data-acquisition system include that company information data obtains subsystem and Company Knowledge production
Weigh information acquisition subsystem；
Step 2: carrying out multi-source data using data-acquisition system and obtain company information and intellectual property information；
Step 3: the company information and Intellectual Property Right of Enterprises information of vectorization enterprise respectively；
Step 4: building enterprise's vector characteristics tree；
Step 5: building circular recursion neural network model；
Step 6: building data set, trained and assessments recurrent neural network become final assessment network；
Step 7: using data-acquisition system as data input port, assess network as arithmetic center and form assessment models.
2. a kind of construction method of Intellectual Property Right of Enterprises appraisal Model according to claim 1, it is characterised in that: institute
It states in step 1 company information data and obtains subsystem and connect with industrial and commercial bureau enterprise information system, Intellectual Property Right of Enterprises acquisition of information
Subsystem is connect with State Intellectual Property Office's patent information system, and the company information data obtains subsystem and Company Knowledge produces
Power information acquisition subsystem includes server configuration and data extractor, and the server configuration includes:
Parameter setting provides the setting of the parameter configuration of data extractor node by a configuration server, and in data
When grabber sends parameter-configuring request to server, matching for the data extractor is returned to according to the unique number of data extractor
Confidence breath；
Point extension, when increasing data extractor node, configuration server receives the node and is transmitted across by socket agreement
The order of the addition data extractor node come, configuration server add a record into configuration information data, and will operation
Point extension: when increasing data extractor node, configuration server receives the node and is sended over by socket agreement
Addition data extractor node order, configuration server adds a record into configuration information data, and by operation
The sum of data extractor node adds 1；
Abnormal monitoring: each data extractor node, which is sent every the 1h time to configuration server, indicates this data extractor node
The order operated normally, configuration server record each data extractor node in data extractor node state list
ID and the corresponding node that finally receives issue the time of order for indicating to operate normally；Configuration server 1h time meeting
Ergodic data grabber node state list, if some data extractor node finally issues the life for indicating to operate normally
The time interval of current time is enabled to be greater than 15 minutes, then it represents that exception has occurred in the data extractor, and configuration server will transport
The sum of the sum of capable grabber subtracts 1 node adds 1；
Abnormal monitoring: each data extractor node every 1h time sends to configuration server is indicating this data extractor node just
In the order of normal operation, configuration server records each data extractor node in data extractor node state list
ID and the corresponding node that finally receives issue the time for indicating the order operated normally；Configuration server is every the 1h time
Meeting ergodic data grabber node state list, if some grabber node finally issues the order for indicating to operate normally
Time interval to current time is greater than 15 minutes, then it represents that exception has occurred in the data extractor, and configuration server will be run
The sum of grabber subtract 1；
Load balancing: the task that each data extractor is completed is identical, and the load on data extractor refers to configuration server
The number that grab number of videos is assigned on each data extractor；Load balancing is related to two stages；First stage
For the distribution of data, second stage is opened after primary data grabber is completed by what the data extractor recorded grabbed
Beginning and end time calculate all grabbers and complete the times required for primary grab, when the next view for having new needs to grab
Frequency is when being added in list, according to ratio of the crawl under each grabber last registration the time required to primary by these new numbers
According to being assigned to each grabber node；
Data extractor includes link grabber, web-page parser, data storage and updates controller, links grabber, presses
Industrial and commercial bureau's enterprise information system is scanned according to breadth traversal algorithm, obtain meet condition webpage link, and by these
Link storage is in a linked database；Meanwhile the mark whether being accessed for link addition each in database, if some
By one of node visit mistake, then other then skip the node, next link are obtained, if this link is not interviewed for link
It asks, then access the page and the link for belonging to this website for including in the webpage is added in linked database；The satisfaction
It is required that webpage refer to containing enterprise name, social Unicode, address and business scope webpage；
Web-page parser, obtains the link of all satisfactions from data storage, and load links corresponding webpage, then carries out net
The parsing of page, obtains required company information；Analytic method library is constructed, if the method that parsing is not indicated in link, traverses
Each method parses each link, and when link is successfully parsed, the analytic method of the link is labeled as this method, if
When being linked at using this method parsing failure of analytic method has been designated, then has added new parsing side into analytic method library
Method；
Data storage, for storing company information lists of links, enterprise's essential information and statistical data, analytic method library number
According to；
The update controller makes according to the renewal frequency of every link in the company information lists of links stored in database
With the frequency of timed task control web-page parser analyzing web page.
3. a kind of construction method of Intellectual Property Right of Enterprises appraisal Model according to claim 1, it is characterised in that: institute
State the detailed process of step 2 are as follows:
In same administrative division, company information data acquisition subsystem connect acquisition enterprise's base with industrial and commercial bureau's enterprise information system
This information, Intellectual Property Right of Enterprises information acquisition subsystem enterprise corresponding to the connection acquisition of State Intellectual Property Office's patent information system
Intellectual property information, intellectual property information includes patent, trade mark, copyright, integrated circuit and new variety of plant；
For multi-source data information, by the mark of society, affiliated enterprise Unicode, by knowing in company information and enterprise
Know property information to be polymerize as unit of enterprise.
4. a kind of construction method of Intellectual Property Right of Enterprises appraisal Model according to claim 3, it is characterised in that: institute
State the detailed process in step 3 are as follows:
Intellectual property evaluation expertise and Feature Engineering method are introduced, data cleansing rule is established, to business data information
It is screened and is identified, rejected garbage, merge redundancy, mark important information；
The company information and Intellectual Property Right of Enterprises information for distinguishing enterprise establish vectorization rule, realize vectorization, have specifically included
It is filled up and the normalized work of data at dirty data processing, missing values, wherein to the middle vectorization result a of any object
() indicates.
5. a kind of construction method of Intellectual Property Right of Enterprises appraisal Model according to claim 3, it is characterised in that: institute
State the detailed process in step 4 are as follows:
By the company information of enterprise and Intellectual Property Right of Enterprises information vector as a result, root node as enterprise vector characteristics tree,
Wherein root node is expressed as r, and poor household's vector characteristics tree table is shown as Tr, the vectorization result expression of root node are as follows:
ra=a (r)；
In enterprise vector characteristics tree TrIn, single intermediate node, i.e. vector characteristics tree T are constructed according to intellectual property classificationrIn
The vectorization of any intermediate node p indicates paBy the y that has the rightaWith failure saTwo parts composition is the intellectual property of authorization to application,
With the filling node q for not including any informationaInstead of, then, point pa、ya、sa、qaIt indicates are as follows:
pa=[ya: qa]
ya=a (y)
qa=a (q)
sa=[0,0 ... ..0,0]
According to the actual production relationship between Intellectual Property Right of Enterprises in reality production, structure enterprise vector characteristics tree.
6. a kind of construction method of Intellectual Property Right of Enterprises appraisal Model according to claim 5, it is characterised in that: institute
State the detailed process in step 5 are as follows:
Deep learning model is circular recursion neural network model M, and using the recursive nature of tree, the vector of tree is indicated by its root section
It puts and indicates to generate by nonlinear operation using its child node as the vector of the subtree collection of root, wherein appoint in vector characteristics tree
The vectorization of one node indicates, is provided by step 4, and the vector of subtree collection is indicated by the way that subtree is sequentially input shot and long term memory
Layer (LSTM), which calculates, to be generated, formally, it is assumed that enterprise vector characteristics tree Tr=(V, E), wherein V and E respectively represents the section of tree
Point set and line set.Further, it is p to any node in V, remembers that the child node collection of p is combined into C={ c1, c2..., ci...,
c|c|, and corresponding subtree collection isWherein ciForRoot node, then be with p
The vectorization of the subtree of root indicates calculation formula are as follows:
Encode (T)=σ (Wparent*p-encoding+Wsubtrees*Encode(Fr))
Wherein, σ () indicates activation primitive, Wparent、WsubtreesParameter, p-encoding be the vectorization of node p as a result,
Encode (F) is that the vectorization of subtree indicates to sequentially input LSTM layers of final output, is indicated are as follows:
It particularly points out, since the root node r of vector characteristics tree indicates user data information, in specific calculate, recursive operation
Terminal calculation formula are as follows:
Encode(Tr)=σ (Wroot*r_encoding+Wsubtrees*Encode(Tr))
Wherein, WrootIt is the known parameters in addition set, r-encoding is the vectorization of root node r as a result, Encode (Tr) be
The subtree collection vectorization of r as a result,
By Encode (Tr) input circular recursion neural network model in subsequent full articulamentum determined.
7. a kind of construction method of Intellectual Property Right of Enterprises appraisal Model according to claim 6, it is characterised in that: institute
State the detailed process in step 6 are as follows:
Data set D is constructed based on company information vector characteristics tree, wherein evaluation quantity is the valuation of the existing intellectual property of enterprise, is commented
Estimating is enterprise's vector characteristics tree, the valuation without no longer valid intellectual property, and divides training set and test set；
Using training set D training circular recursion neural network M, and in test central evaluation M effect；
Circular recursion neural network M after training is put into real enterprise intellectual property value evaluation work and is used.
8. a kind of construction method of Intellectual Property Right of Enterprises appraisal Model according to claim 7, it is characterised in that: institute
State the detailed process in step 7 are as follows:
Use data-acquisition system as data input port, assesses network as arithmetic center and form assessment models, obtained in data
System is taken to belong to enterprise's identification information, enterprise name or social Unicode, data-acquisition system obtain the basic letter of enterprise
The A to Z of property information of data and enterprise is ceased, and is put into assessment network and carries out generation assessment result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910194965.1A CN109933703A (en) | 2019-03-14 | 2019-03-14 | A kind of construction method of Intellectual Property Right of Enterprises appraisal Model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910194965.1A CN109933703A (en) | 2019-03-14 | 2019-03-14 | A kind of construction method of Intellectual Property Right of Enterprises appraisal Model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109933703A true CN109933703A (en) | 2019-06-25 |
Family
ID=66987252
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910194965.1A Withdrawn CN109933703A (en) | 2019-03-14 | 2019-03-14 | A kind of construction method of Intellectual Property Right of Enterprises appraisal Model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109933703A (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110991441A (en) * | 2019-12-13 | 2020-04-10 | 王文斌 | Asset assessment method and device based on image recognition and computer storage medium |
| CN111160783A (en) * | 2019-12-30 | 2020-05-15 | 北京阿尔山区块链联盟科技有限公司 | Method and system for evaluating digital asset value and electronic equipment |
| CN111382843A (en) * | 2020-03-06 | 2020-07-07 | 浙江网商银行股份有限公司 | Method and device for establishing upstream and downstream relation recognition model of enterprise and relation mining |
| CN111566691A (en) * | 2019-09-20 | 2020-08-21 | 钟山 | Intellectual property value management and operation method, device, medium and computing equipment |
| CN112965989A (en) * | 2021-03-04 | 2021-06-15 | 浪潮云信息技术股份公司 | Main body scattered data query and research and judgment method |
| CN113112380A (en) * | 2021-05-12 | 2021-07-13 | 北京大学 | Intellectual property service value evaluation system |
| CN113191870A (en) * | 2021-01-19 | 2021-07-30 | 迅鳐成都科技有限公司 | Intellectual property value evaluation method and system based on block chain |
| CN114974593A (en) * | 2022-07-11 | 2022-08-30 | 江苏优创生物医学科技有限公司 | Enterprise health management evaluation method and system based on block chain |
| CN117575827B (en) * | 2024-01-16 | 2024-05-03 | 之江实验室科技控股有限公司 | Intelligent visual management system and method for enterprise report |
-
2019
- 2019-03-14 CN CN201910194965.1A patent/CN109933703A/en not_active Withdrawn
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111566691A (en) * | 2019-09-20 | 2020-08-21 | 钟山 | Intellectual property value management and operation method, device, medium and computing equipment |
| CN110991441A (en) * | 2019-12-13 | 2020-04-10 | 王文斌 | Asset assessment method and device based on image recognition and computer storage medium |
| CN111160783A (en) * | 2019-12-30 | 2020-05-15 | 北京阿尔山区块链联盟科技有限公司 | Method and system for evaluating digital asset value and electronic equipment |
| CN111160783B (en) * | 2019-12-30 | 2023-10-24 | 北京阿尔山区块链联盟科技有限公司 | Digital asset value evaluation method and system and electronic equipment |
| CN111382843A (en) * | 2020-03-06 | 2020-07-07 | 浙江网商银行股份有限公司 | Method and device for establishing upstream and downstream relation recognition model of enterprise and relation mining |
| CN111382843B (en) * | 2020-03-06 | 2023-10-20 | 浙江网商银行股份有限公司 | Method and device for establishing enterprise upstream and downstream relationship identification model and mining relationship |
| CN113191870A (en) * | 2021-01-19 | 2021-07-30 | 迅鳐成都科技有限公司 | Intellectual property value evaluation method and system based on block chain |
| CN113191870B (en) * | 2021-01-19 | 2023-08-08 | 迅鳐成都科技有限公司 | Intellectual property value evaluation method and system based on blockchain |
| CN112965989A (en) * | 2021-03-04 | 2021-06-15 | 浪潮云信息技术股份公司 | Main body scattered data query and research and judgment method |
| CN113112380A (en) * | 2021-05-12 | 2021-07-13 | 北京大学 | Intellectual property service value evaluation system |
| CN114974593A (en) * | 2022-07-11 | 2022-08-30 | 江苏优创生物医学科技有限公司 | Enterprise health management evaluation method and system based on block chain |
| CN117575827B (en) * | 2024-01-16 | 2024-05-03 | 之江实验室科技控股有限公司 | Intelligent visual management system and method for enterprise report |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109933703A (en) | A kind of construction method of Intellectual Property Right of Enterprises appraisal Model | |
| Schoder | Forecasting the success of telecommunication services in the presence of network effects | |
| US20050182739A1 (en) | Implementing data quality using rule based and knowledge engineering | |
| Van der Aalst | Trends in business process analysis | |
| US7353230B2 (en) | Dynamic distributed customer issue analysis | |
| Hsiao et al. | First‐mover strategy, resource capacity alignment, and new product performance: a framework for mediation and moderation effects | |
| CN109388697A (en) | Handle the method, apparatus and electric terminal of question and answer corpus | |
| CN111401722B (en) | Intelligent decision method and intelligent decision system | |
| Karasoy | How do remittances to the Philippines affect its environmental sustainability? Evidence based on the augmented ARDL approach | |
| CN110286938A (en) | For exporting the method and apparatus for being directed to the evaluation information of user | |
| Le Dinh et al. | Towards a knowledge-based framework for enterprise content management | |
| CN112837140A (en) | Data processing method, device, equipment and storage medium | |
| Ostapchenko et al. | Functional organization of system of support of decision-making of organizational management | |
| CN110276593A (en) | Object recommendation method, apparatus, server and storage medium | |
| Farajpour et al. | Information sharing assessment in supply chain: hierarchical fuzzy rule-based system | |
| CN109522292A (en) | Data processing equipment and method based on power grid standard uniform information model | |
| Hsu | Organizational innovation strategies: The value co-creation strategy (VCS) model | |
| CN107767278B (en) | Method and device for constructing community hierarchy | |
| CN108711074A (en) | Business sorting technique, device, server and readable storage medium storing program for executing | |
| Baida et al. | E-business cases assessment: From business value to system feasibility | |
| Lu et al. | Fuzzy synthetic evaluation on customer loyalty based on analytic hierarchy process | |
| Akal et al. | Challenges of identifying and utilizing big data analytics in a resource-constrained environment: In the case of Ethiopia | |
| Ganzarain et al. | Testing successful Business Model using System Dynamics | |
| Oyoo | Collaboration-Based Automatic Data Validation Framework for Enterprise Asset Management | |
| CN116862260A (en) | Method and device for determining implementation level of software project, electronic equipment and medium |
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 | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20190625 |
|
| WW01 | Invention patent application withdrawn after publication |