CN108256552A - Common people close friend's index assessment method and system based on big data sorting algorithm - Google Patents
Common people close friend's index assessment method and system based on big data sorting algorithm Download PDFInfo
- Publication number
- CN108256552A CN108256552A CN201711366824.0A CN201711366824A CN108256552A CN 108256552 A CN108256552 A CN 108256552A CN 201711366824 A CN201711366824 A CN 201711366824A CN 108256552 A CN108256552 A CN 108256552A
- Authority
- CN
- China
- Prior art keywords
- work order
- business
- determined
- common people
- close friend
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- 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
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
Abstract
The invention discloses common people close friend's index assessment method and system based on big data sorting algorithm, this method includes:Obtain work order to be determined;The first business datum and the first emotion lexical data are extracted from work order to be determined;According to the first business datum, the first emotion lexical data, business scope score model and mood grade separation model, the first scoring rate corresponding to work order to be determined is calculated;First scoring rate is input in SVM classifier and carries out data processing, exports common people close friend's index.One system includes acquiring unit, extraction unit and first processing units to fourth processing unit.Another system includes memory and the processor for performing above-mentioned common people close friend's index assessment method based on big data sorting algorithm.The present invention can quickly and accurately evaluate common people close friend's index corresponding to work order, reduce operating mistake and operative error caused by manual sort.The method and system of the present invention can be widely applied in friendly index evaluation field.
Description
Technical field
The present invention relates to big data treatment technology more particularly to a kind of common people close friend's indexes based on big data sorting algorithm
Assessment method and system.
Background technology
Technology word is explained:
Big data sorting algorithm:It refers to based on support vector machines, neural network etc., to find out one in database
The common feature of group data object and the algorithm that different classes is divided into according to classification mode;The purpose is to pass through mould of classifying
Type, will be in the maps data items in database to some given classification.
Support vector machines:In machine learning, support vector machines (SVM goes back support vector network) is and relevant study
The related supervised learning model of algorithm can analyze data, recognition mode, for classification and regression analysis;Give one group of training
Sample, each label is two classes, and a SVM training algorithm is established to obtain a model, distribute new example to be a kind of or
Other classes become non-probability binary linearity classification.
Expert analysis mode model:So-called scoring is exactly to make measurement to certain attributes of things or influence;Its essence is main bodys
(estimator or scoring expert) is to the understanding of object (evaluation object) essential attribute and the rule of development;And the process evaluated is evaluation
Person is according to the awareness to object and the estimator level of understanding, values and psychological factor in itself to the attribute of evaluation object
The process described.The bridge and tie for contacting subject and object are to compare, i.e., with certain determining standard and evaluation object
It compares, the cardinal principle and means of expert analysis mode are also to compare.
With the fast development of social economy in recent years, 12345 government affairs service hotlines accept part consumer inquiries, complaint,
Report, opinion and suggestion, and for the content that these accept are that broad masses of the people deliver to one's home and come " investigation ",
It is the big data that department in charge of industry and commerce's research is strengthened and improved market surpervision work and must refer to.Therefore, to 12345
Work order data are analyzed and processed, this is to realize that market " big data " is effectively explored and important way to what market surpervision ability converted
Diameter.
12345 government affairs service hotline center of the department in charge of industry and commerce is directly facing consumers in general, close to the people,
It is close to the market, can all generate mass data information daily, therefore, the common people established using the excavation of these data are friendly
Index can reflect the Variation Features and rule of common people's mood, reflection main market players honest operation situation, quotient accurately and in time
Product and service quality condition and market fair deal order situation, even more evaluate department in charge of industry and commerce's market surpervision into
" main examiner " of effect.However, at present for 12345 work order processing, be only:First 12345 work order content is divided
Class then according to worksheet object, distributes to relevant departments and law enforcement is gone to solve.As it can be seen that for existing 12345 work order data
Processing mode has lacked the analysis to common people's mood, masses' satisfaction, that is, has lacked and the common people close friend of 12345 work orders is referred to
Several evaluations, so often can not the accurately overall common people's emotional state in area from terms of macroscopic view.In addition, 12345 works
Single data volume is very big, therefore, how rapidly and accurately to calculate evaluation and obtains common people close friend's index, this is that there is an urgent need to solve at present
Certainly the technical issues of.
Invention content
In order to solve the above-mentioned technical problem, the object of the present invention is to provide a kind of common people friends based on big data sorting algorithm
Good index assessment method and system can rapidly and accurately assess common people close friend's index.
First technical solution of the present invention is:A kind of common people close friend's index evaluation based on big data sorting algorithm
Method includes the following steps:
Obtain work order to be determined;
The first business datum and the first emotion lexical data are extracted from the work order to be determined acquired;
According to the first business datum extracted, the business scope belonging to work order to be determined is judged, then basis is waited to sentence
Determine the business scope belonging to work order and business scope score model, acquire the significance level corresponding to work order to be determined
Value;
According to the first business datum extracted, the first emotion lexical data and mood grade separation model, judge
Mood number of degrees belonging to work order to be determined;
Importance value and affiliated mood number of degrees according to corresponding to work order to be determined, are calculated work to be determined
The first scoring rate corresponding to list;
The first scoring rate corresponding to work order to be determined is input in SVM classifier and carries out data processing, so as to export
Common people close friend's index.
Second technical solution of the present invention is:A kind of common people close friend's index evaluation based on big data sorting algorithm
System, including:
Acquiring unit, for obtaining work order to be determined;
Extraction unit, for extracting the first business datum and the first emotion vocabulary from the work order to be determined acquired
Data;
First processing units, for according to the first business datum extracted, judging the business belonging to work order to be determined
Field, then the business scope according to belonging to work order to be determined and business scope score model, acquire work order to be determined
Corresponding importance value;
Second processing unit, for according to the first business datum, the first emotion lexical data and mood extracted etc.
Grade disaggregated model, judges the mood number of degrees belonging to work order to be determined;
Third processing unit, for the importance value according to corresponding to work order to be determined and affiliated mood grade
Number, is calculated the first scoring rate corresponding to work order to be determined;
Fourth processing unit carries out for the first scoring rate corresponding to work order to be determined to be input in SVM classifier
Data processing, so as to export common people close friend's index.
Third technical solution of the present invention is:A kind of common people close friend's index evaluation based on big data sorting algorithm
System, including:
At least one processor;
At least one processor, for storing at least one program;
When at least one program is performed by least one processor so that at least one processor is realized
Common people close friend's index assessment method based on big data sorting algorithm as described in the first technical solution.
The advantageous effect of the method for the present invention and system is:It, can quickly, accurately by using the method and system of the present invention
Common people close friend's index corresponding to work order is evaluated on ground, substantially without manual operation is related to, is substantially reduced manual sort and is made
Into operating mistake and operative error, mitigate the workload and work load of staff, enable evaluation treatment effeciency and
Accuracy is greatly improved.
Description of the drawings
Fig. 1 is a kind of step flow signal of common people close friend's index assessment method based on big data sorting algorithm of the present invention
Figure;
Fig. 2 is a kind of first structure frame of common people close friend's index assessment system based on big data sorting algorithm of the present invention
Figure;
Fig. 3 is a kind of second structural frames of common people close friend's index assessment system based on big data sorting algorithm of the present invention
Figure;
Fig. 4 is an a kind of specific embodiment of common people close friend's index assessment method based on big data sorting algorithm of the present invention
Steps flow chart schematic diagram;
Fig. 5 is mood grade separation schematic diagram.
Specific embodiment
The present invention is described in further detail in the following with reference to the drawings and specific embodiments.In for the examples below
Number of steps is set only for the purposes of illustrating explanation, and the sequence between step does not do any restriction, each in embodiment
The execution sequence of step can be adaptively adjusted according to the understanding of those skilled in the art.
As shown in Figure 1, a kind of common people close friend's index assessment method based on big data sorting algorithm is present embodiments provided,
Include the following steps:
Obtain work order to be determined;
The first business datum and the first emotion lexical data are extracted from the work order to be determined acquired;
According to the first business datum extracted, the business scope belonging to work order to be determined is judged, then basis is waited to sentence
Determine the business scope belonging to work order and business scope score model, acquire the significance level corresponding to work order to be determined
Value;
According to the first business datum extracted, the first emotion lexical data and mood grade separation model, judge
Mood number of degrees belonging to work order to be determined;
Importance value and affiliated mood number of degrees according to corresponding to work order to be determined, are calculated work to be determined
The first scoring rate corresponding to list;
The first scoring rate corresponding to work order to be determined is input in SVM classifier and carries out data processing, so as to export
Common people close friend's index.
The preferred embodiment of this method is further used as, first scoring rate by corresponding to work order to be determined inputs
Data processing is carried out into SVM classifier, the step for so as to export common people close friend's index before be equipped with structure SVM classifier this
The step for one step, the structure SVM classifier, includes:
Obtain multiple sample work orders;
Using the second business datum extracted from multiple sample work orders and the second emotion lexical data, structure obtains feelings
Thread grade separation model;
Using the importance value corresponding to multiple business scopes, structure obtains business scope score model;
Using mood grade separation model and business scope score model, second corresponding to multiple sample work orders is calculated
Scoring rate;
Mark common people close friend's index corresponding to multiple sample work orders;
The second scoring rate corresponding to multiple sample work orders is respectively constituted into trained input data set and test input data
Collection;
By common people close friend's index composing training output data set corresponding to multiple sample work orders and test output data set;
A parameter is chosen to coming respectively as error penalty factor initial value and nuclear parameter from preset multiple parameters centering
Initial value;
Using training input data set, training output data set, error penalty factor initial value and nuclear parameter initial value come
SVM classifier is trained;
Using input data set and test output data set is tested, the SVM classifier institute obtained after training of judgement is defeated
Whether the common people close friend's index gone out meets preset accuracy rate requirement, if so, the svm classifier that will be obtained after current training
SVM classifier of the device as required structure;Described a ginseng is chosen conversely, then returning to re-execute from preset multiple parameters centering
It is several to coming respectively as error penalty factor initial value and nuclear parameter initial value the step for.
As it can be seen that the business datum and emotion lexical data that are extracted from sample work order be referred to as the second business datum and
Second emotion lexical data;The business datum and emotion lexical data extracted from work order to be determined is referred to as the first business
Data and the first emotion lexical data;Scoring rate corresponding to sample work order is known as the second scoring rate;Corresponding to work order to be determined
Scoring rate be known as the first scoring rate.
The preferred embodiment of this method is further used as, it is described to utilize the second industry extracted from multiple sample work orders
It the step for data of being engaged in and the second emotion lexical data, structure obtains mood grade separation model, specifically includes:
The second business datum and the second emotion lexical data are extracted from multiple sample work orders;
The the second emotion lexical data extracted is compared with the emotion lexical types in emotion lexicon, so as to
Go out the emotion lexical types belonging to the second emotion lexical data;
The second business datum extracted and the service conditions type in business library are compared, so as to obtain the second industry
Service conditions type belonging to data of being engaged in;
By the emotion lexical types belonging to the second emotion lexical data and the service conditions type belonging to the second business datum
Form the feature of sample work order;
According to the influence degree to common people close friend's index, mood grade separation is carried out to the feature of sample work order, so as to
Go out the mood number of degrees corresponding to the feature of sample work order;
Mapping relations between the feature of sample work order and corresponding mood number of degrees are formed into mood grade separation model.
The preferred embodiment of this method is further used as, utilizes the importance value corresponding to multiple business scopes, structure
The step for building to obtain business scope score model, specifically includes:
Calculate the field base score value corresponding to each business scope;
Field base score value according to corresponding to each business scope, calculates the important journey corresponding to each business scope
Angle value;
Mapping relations between business scope and corresponding importance value are formed into business scope score model.
The preferred embodiment of this method is further used as, the calculating of the field base score value corresponding to the business scope is public
Formula is as follows:
bi(j)(k)=m-xi(j)(k)+1
Wherein, bi(j)(k) when being expressed as k-th of expert i-th of business scope being come jth position, i-th of business scope institute
Corresponding field base score value;xi(j)(k) when representing that i-th of business scope is come jth position by k-th of expert, i-th of business scope
Corresponding score value;M is expressed as the total number of business scope.
The preferred embodiment of this method is further used as, the calculating of the importance value corresponding to the business scope is public
Formula is as follows:
Wherein, SiIt is expressed as the importance value corresponding to i-th of business scope;N is expressed as expert's total number of persons;bi(j)Table
When being shown as i-th of business scope and coming jth position, the field base score value corresponding to i-th of business scope;NjIt represents i-th of industry
Business field comes expert's number of jth position.
Be further used as the preferred embodiment of this method, the importance value according to corresponding to work order to be determined with
And affiliated mood number of degrees, the step for the first scoring rate corresponding to work order to be determined is calculated, it is specially:
After importance value corresponding to work order to be determined is multiplied with affiliated mood number of degrees, obtained product
As the first scoring rate corresponding to work order to be determined.
As shown in Fig. 2, the present embodiment additionally provides a kind of common people close friend's index evaluation system based on big data sorting algorithm
System, including:
Acquiring unit, for obtaining work order to be determined;
Extraction unit, for extracting the first business datum and the first emotion vocabulary from the work order to be determined acquired
Data;
First processing units, for according to the first business datum extracted, judging the business belonging to work order to be determined
Field, then the business scope according to belonging to work order to be determined and business scope score model, acquire work order to be determined
Corresponding importance value;
Second processing unit, for according to the first business datum, the first emotion lexical data and mood extracted etc.
Grade disaggregated model, judges the mood number of degrees belonging to work order to be determined;
Third processing unit, for the importance value according to corresponding to work order to be determined and affiliated mood grade
Number, is calculated the first scoring rate corresponding to work order to be determined;
Fourth processing unit carries out for the first scoring rate corresponding to work order to be determined to be input in SVM classifier
Data processing, so as to export common people close friend's index.
The preferred embodiment of this system is further used as, further includes the construction unit for building SVM classifier, it is described
Construction unit includes:
Acquisition module, for obtaining multiple sample work orders;
First processing module, for utilizing the second business datum and the second emotion word extracted from multiple sample work orders
Remittance data, structure obtain mood grade separation model;
Second processing module, for using the importance value corresponding to multiple business scopes, structure to obtain business scope
Score model;
Third processing module for utilizing mood grade separation model and business scope score model, calculates multiple samples
The second scoring rate corresponding to this work order;
Fourth processing module, for marking common people close friend's index corresponding to multiple sample work orders;
5th processing module, for the second scoring rate corresponding to multiple sample work orders to be respectively constituted trained input data
Collection and test input data set;
6th processing module, for by common people close friend's index composing training output data set corresponding to multiple sample work orders
With test output data set;
7th processing module, for from preset multiple parameters centering choose a parameter to come respectively as error punishment because
Sub- initial value and nuclear parameter initial value;
8th processing module, for utilizing training input data set, training output data set, error penalty factor initial value
SVM classifier is trained with nuclear parameter initial value;
9th processing module, for using input data set and test output data set is tested, being obtained after training of judgement
To common people close friend's index for being exported of SVM classifier whether meet preset accuracy rate requirement, if so, knot will be trained currently
SVM classifier of the SVM classifier obtained after beam as required structure;The 7th processing module institute is re-executed conversely, then returning
Corresponding flow chart of data processing.
As shown in figure 3, the present embodiment additionally provides a kind of common people close friend's index evaluation system based on big data sorting algorithm
System, including:
At least one processor;
At least one processor, for storing at least one program;
When at least one program is performed by least one processor so that at least one processor is realized
Common people close friend's index assessment method based on big data sorting algorithm as described in above-mentioned embodiment of the method.
It is further elaborated below in conjunction with particular preferred embodiment to be done to the present invention.
In order to realize the fast and accurately evaluation of common people close friend's index, the present invention using big data sorting algorithm technology,
The technologies such as SVM vector machines and expert analysis mode, the 12345 work order data based on magnanimity establish 12345 common people close friend's index score moulds
Type extracts work order content, corresponding scoring content, more scientific, accurately progress common people close friend index evaluation.
As shown in figure 4, a kind of common people close friend's index assessment method based on big data sorting algorithm, the step specifically included
It is rapid as follows.
Step 1: structure SVM classifier.Specifically, the step 1 preferably includes following sub-step.
S101, multiple sample work orders are obtained.
Specifically, 12345 work orders of magnanimity are obtained as required sample work order.
S102, the second business datum extracted from multiple sample work orders and the second emotion lexical data, structure are utilized
Obtain mood grade separation model.
Specifically, the step S102 preferably includes following sub-step.
S1021, the second business datum and the second emotion lexical data are extracted from multiple sample work orders.
Specifically, from each sample work order, corresponding business datum and emotion lexical data are extracted.
S1022, the second emotion lexical data extracted is compared with the emotion lexical types in emotion lexicon,
So as to obtain the emotion lexical types belonging to the second emotion lexical data.
Specifically, corresponding emotion lexical data will be extracted from each sample work order, and in emotion lexicon
Emotion lexical types are compared, and just emotional semantic classification can be carried out to the emotion lexical data in this way, so as to identify the emotion word
The emotion lexical types converged belonging to data.
S1023, the second business datum extracted and the service conditions type in business library are compared, so as to obtain
Service conditions type belonging to second business datum.
Specifically, corresponding business datum will be extracted from each sample work order, with the service conditions in business library
Type is compared, and just situation classification can be carried out to the business datum in this way, so as to obtain the business feelings belonging to the business datum
Condition type.
S1024, by the emotion lexical types belonging to the second emotion lexical data and the business feelings belonging to the second business datum
Condition type forms the feature of sample work order.
It specifically, will be in the emotion lexical types belonging to the emotion lexical data in sample work order and sample work order
Service conditions type belonging to business datum forms the work order feature corresponding to the sample work order.
S1025, according to the influence degree to common people close friend's index, mood grade separation is carried out to the feature of sample work order,
So as to obtain the mood number of degrees corresponding to the feature of sample work order to get to the mood number of degrees corresponding to sample work order.
Specifically, in the present embodiment, according to the influence degree to common people close friend's index, using clustering technique to sample work
Single feature carries out mood grade separation, is divided into 5 grades, as shown in figure 5, level-one (high-risk), two level (danger), three-level respectively
(general to complain), level Four (complaining tendency) and Pyatyi (no to complain tendency), that is to say, that each mood grade, which can correspond to, to be included
There is the feature of at least one sample work order.
S1026, the mapping relations between the feature of sample work order and corresponding mood number of degrees are formed into mood ranking score
Class model.
Specifically, by the feature of the feature of sample work order and sample work order corresponding to mood number of degrees, between the two
Mapping relations form mood grade separation model.
S103, the importance value corresponding to using multiple business scopes, structure obtain business scope score model.
In the present embodiment, (but business scope is not limited to as shown in table 1 for the number of business scope and concrete type
Number and type shown in table 1);In this step, the importance value corresponding to each business scope is mainly calculated,
So as to build to obtain required business scope score model.Specifically, the table 1 is as follows.
Table 1
Specifically, the step S103 preferably includes following sub-step.
S1031, field base score value corresponding to each business scope is calculated.
In the present embodiment, first, expert analysis mode model is applied in the ranking of business scope, that is, utilize expert analysis mode
Model ranks multiple business scopes;
Assuming that the evaluation field (i.e. above-mentioned business scope) involved by common people close friend's index shares m, n expert is shared
Evaluation is participated in, i.e., n expert respectively evaluates m business scope, at this point, setting the scoring value set that a certain expert k is provided
For Xi (j)(k), this score value set expression expert K (i.e. k-th of expert) m field is carried out ranking as a result, and this
One set Xi (j)(k)Included in element xi(j)(k) be then expressed as kth (k=1,2, n) a expert is by the i-th (i=
1,2, m) a business scope is when coming jth position, the score value corresponding to i-th of business scope;J is expressed as ranking position
Number;
In the present embodiment, the numerical value of the score value be ranking digit, for example, the 5th expert to m business scope into
The result of row ranking is x1(2) (5)、x2(4) (5)、……、xm(15) (5), therefore it can be seen that in the scoring given by the 5th expert
In value set, the 1st business scope come the 2nd, the 2nd business scope come the 4th ..., m-th of business scope come
15th, at this point, the score value x corresponding to the 1st business scope1(2) (5)For the score value corresponding to 2, the 2nd business scopes
x2(4) (5)For 4 ..., the score value x corresponding to m-th of business scopem(15) (5)It is 15;
Then, the scoring value set provided respectively based on each expert (common n expert) calculates each business neck
Field base score value b corresponding to domaini(j)(k);
Specifically, this step is mainly by each scoring value set Xi (j)(k)Be converted to corresponding base score value set Bi (j
)(k), the base score value, which show certain field on the position of certain ranking there are one basis point, wherein, influence degree is bigger
Field, base score value is then bigger, to ensure the influence power of key areas.
Preferably, the calculation formula of the field base score value corresponding to the business scope is as follows:
bi(j)()=m-xi(j)(k)+1
Wherein, bi(j)(k) when being expressed as k-th of expert i-th of business scope being come jth position, i-th of business scope institute
Corresponding field base score value;xi(j)(k) when representing that i-th of business scope is come jth position by k-th of expert, i-th of business scope
Corresponding score value;M is expressed as the total number of business scope.
S1032, the field base score value according to corresponding to each business scope, calculate corresponding to each business scope
Importance value.
Preferably, the calculation formula of the importance value corresponding to the business scope is as follows:
Wherein, SiIt is expressed as the importance value corresponding to i-th of business scope;N is expressed as expert's total number of persons;bi(j)Table
When being shown as i-th of business scope and coming jth position, the field base score value corresponding to i-th of business scope;NjIt represents i-th of industry
Business field comes expert's number of jth position, is also equivalent to, NjIt is expressed as the number that a certain business scope is approved of to come jth position.
S1033, the mapping relations between business scope and corresponding importance value are formed into business scope score model.
Specifically, by business scope and the importance value corresponding to business scope, mapping relations between the two are formed
Business scope score model.
S104, using mood grade separation model and business scope score model, calculate corresponding to multiple sample work orders
The second scoring rate.
Specifically, for the scoring rate corresponding to work order, specific calculation is, by the significance level corresponding to work order
After value is multiplied with affiliated mood number of degrees, obtained product is as the scoring rate corresponding to work order, therefore, for work order
Corresponding scoring rate, the calculation formula preferably used are as follows:
Vw=Si*VEmotion
Wherein, SiBe expressed as the business scope belonging to work order, corresponding to importance value;VEmotionIt is expressed as work order institute
Corresponding mood number of degrees;
It is according to mood corresponding to sample work order etc. as it can be seen that for the second scoring rate corresponding to sample work order
The importance value corresponding to business scope belonging to series and sample work order, the two obtained value after being multiplied.
It is right to mark each sample work order institute for common people close friend's index corresponding to S105, the multiple sample work orders of label
The common people close friend's index answered.
S106, the second scoring rate corresponding to multiple sample work orders is respectively constituted to trained input data set and test input
Data set.
S107, common people close friend's index composing training output data set corresponding to multiple sample work orders and test are exported into number
According to collection.
S108, a parameter is chosen to coming respectively as error penalty factor initial value and core from preset multiple parameters centering
Initial parameter value.
Specifically, it is according to the value range of error penalty factor and nuclear parameter γ for the multiple parameter pair
Value range, so as to build the ginseng being made of the numerical value of error penalty factors all in value range and the numerical value of nuclear parameter γ
Obtained from several pairs;Preferably, the value range of the error penalty factor be [1,1000], the value of the nuclear parameter γ
Ranging from [0,100].
S109, using training, input data set, training output data set, error penalty factor initial value and nuclear parameter are initial
Value is trained SVM classifier.
S110, using testing input data set and test output data set, the SVM classifier obtained after training of judgement
Whether the common people close friend's index exported meets preset accuracy rate requirement, if so, the SVM that will be obtained after current training
SVM classifier of the grader as required structure, and keep records of down currently employed error penalty factor and nuclear parameter, i.e.,
(C0,Υ0), that is to say, that it is to (C based on parameter to obtain required SVM classifier at this time0,Υ0) SVM classifier;
Conversely, then returning to re-execute described a parameter is chosen to coming respectively as error from preset multiple parameters centering
The step for penalty factor initial value and nuclear parameter initial value S108, so as to choose next parameter to being punished respectively as error
Penalty factor initial value and nuclear parameter initial value, and combined training input/output data collection, are again trained SVM classifier,
Until common people close friend's index that the SVM classifier obtained after training is exported meets preset accuracy rate requirement.
Step 2: using the SVM classifier built, the evaluation of common people close friend's index is carried out to work order to be determined.
Specifically, the step 2 preferably includes following sub-step.
S201, work order to be determined is obtained.
S202, the first business datum and the first emotion lexical data are extracted from the work order to be determined acquired.
The first business datum that S203, basis extract, judges the business scope belonging to work order to be determined, then basis
Business scope and business scope score model belonging to work order to be determined, acquire the important journey corresponding to work order to be determined
Angle value.
Specifically, according to business scope score model it is found that each business scope have one correspond to importance value, because
This, when judging the business scope belonging to work order to be determined, can wait to sentence using business scope score model to acquire
Determine the importance value corresponding to work order.
The first business datum, the first emotion lexical data and the mood grade separation model that S204, basis extract, sentence
Break the mood number of degrees belonging to work order to be determined.
Specifically, according to mood grade separation model it is found that each sample work order feature correspond to a mood number of degrees,
Therefore, when the feature that the sample work order most like with the feature of work order to be determined is identified from mood grade separation model,
Mood number of degrees corresponding to the feature of sample work order identified as judges the mood number of degrees belonging to work order;
Wherein, for the acquisition modes of the feature of work order to be determined, in shown in above-mentioned steps S1021~S1024
Hold to realize.
S205, the importance value according to corresponding to work order to be determined and affiliated mood number of degrees, are calculated and treat
Judge the first scoring rate corresponding to work order.
Preferably, the step S205 is, by the importance value corresponding to work order to be determined and affiliated mood grade
After number is multiplied, obtained product is as the first scoring rate corresponding to work order to be determined, that is to say, that step S205 is used
This Vw=Si*VEmotionFormula is calculated the first scoring rate corresponding to work order to be determined.
S206, it the first scoring rate corresponding to work order to be determined is input in SVM classifier carries out data processing, so as to
Common people close friend's index is exported, this common people close friend's index exported is that the common people close friend corresponding to required work order to be determined refers to
Number.
It is obtained by above-mentioned, the present invention is using technologies such as big data sorting algorithm technology, SVM vector machines and expert analysis modes come real
The fast and accurately evaluation of common people close friend's index corresponding to existing work order, operating mistake caused by substantially reducing manual sort and
Operative error mitigates the workload and work load of staff, and the treatment effeciency of evaluation and accuracy is enabled to obtain greatly
Raising.And using the building mode of above-mentioned SVM classifier, accurately work order data can be parsed, split out the people
Many emotion degree and business scoring type, common people's emotion degree combination work order service conditions are calculated, the index energy finally obtained
Relevantly reflect the Variation Features and rule of common people's mood, reflection main market players honest operation situation, commodity and service quality
Situation.
In addition, for the method and system of the present invention, it is applicable to government's stability maintenance monitoring system, complaints and denunciation data point
System, 12345 government affairs service hotline systems etc. are studied and judged in analysis.
Content in this preferred embodiment according to the understanding of those skilled in the art can be applicable in after splitting/combining
In 1~embodiment of above-described embodiment 3.
It is that the preferable of the present invention is implemented to be illustrated, but the invention is not limited to the implementation above
Example, those skilled in the art can also make various equivalent variations under the premise of without prejudice to spirit of the invention or replace
It changes, these equivalent deformations or replacement are all contained in the application claim limited range.
Claims (10)
1. a kind of common people close friend's index assessment method based on big data sorting algorithm, which is characterized in that include the following steps:
Obtain work order to be determined;
The first business datum and the first emotion lexical data are extracted from the work order to be determined acquired;
According to the first business datum extracted, the business scope belonging to work order to be determined is judged, then according to work to be determined
Business scope and business scope score model belonging to list, acquire the importance value corresponding to work order to be determined;
According to the first business datum extracted, the first emotion lexical data and mood grade separation model, judge to wait to sentence
Determine the mood number of degrees belonging to work order;
Importance value and affiliated mood number of degrees according to corresponding to work order to be determined, are calculated work order institute to be determined
Corresponding first scoring rate;
The first scoring rate corresponding to work order to be determined is input in SVM classifier and carries out data processing, so as to export the common people
Friendly index.
2. a kind of common people close friend's index assessment method based on big data sorting algorithm, feature exist according to claim 1
In first scoring rate by corresponding to work order to be determined, which is input in SVM classifier, carries out data processing, so as to export the people
The step for the step for structure SVM classifier is equipped with before the step for many close friend's indexes, the structure SVM classifier, wraps
It includes:
Obtain multiple sample work orders;
Using the second business datum extracted from multiple sample work orders and the second emotion lexical data, structure obtains mood etc.
Grade disaggregated model;
Using the importance value corresponding to multiple business scopes, structure obtains business scope score model;
Using mood grade separation model and business scope score model, the second score corresponding to multiple sample work orders is calculated
Rate;
Mark common people close friend's index corresponding to multiple sample work orders;
The second scoring rate corresponding to multiple sample work orders is respectively constituted into trained input data set and test input data set;
By common people close friend's index composing training output data set corresponding to multiple sample work orders and test output data set;
It is initial respectively as error penalty factor initial value and nuclear parameter to coming that a parameter is chosen from preset multiple parameters centering
Value;
Using training input data set, training output data set, error penalty factor initial value and nuclear parameter initial value come to SVM
Grader is trained;
Using testing input data set and test output data set, what the SVM classifier obtained after training of judgement was exported
Whether common people close friend's index meets preset accuracy rate requirement, if so, the SVM classifier obtained after current training is made
SVM classifier for required structure;Described a parameter pair is chosen conversely, then returning to re-execute from preset multiple parameters centering
The step for coming respectively as error penalty factor initial value and nuclear parameter initial value.
3. a kind of common people close friend's index assessment method based on big data sorting algorithm, feature exist according to claim 2
In described using the second business datum and the second emotion lexical data that are extracted from multiple sample work orders, structure obtains feelings
It the step for thread grade separation model, specifically includes:
The second business datum and the second emotion lexical data are extracted from multiple sample work orders;
The the second emotion lexical data extracted is compared with the emotion lexical types in emotion lexicon, so as to obtain
Emotion lexical types belonging to two emotion lexical datas;
The second business datum extracted and the service conditions type in business library are compared, so as to obtain the second business number
According to affiliated service conditions type;
Emotion lexical types belonging to second emotion lexical data and the service conditions type belonging to the second business datum are formed
The feature of sample work order;
According to the influence degree to common people close friend's index, mood grade separation is carried out to the feature of sample work order, so as to obtain sample
Mood number of degrees corresponding to the feature of this work order;
Mapping relations between the feature of sample work order and corresponding mood number of degrees are formed into mood grade separation model.
4. a kind of common people close friend's index assessment method based on big data sorting algorithm, feature exist according to claim 2
In, using the importance value corresponding to multiple business scopes, the step for obtaining business scope score model is built, it is specific
Including:
Calculate the field base score value corresponding to each business scope;
Field base score value according to corresponding to each business scope, calculates the significance level corresponding to each business scope
Value;
Mapping relations between business scope and corresponding importance value are formed into business scope score model.
5. a kind of common people close friend's index assessment method based on big data sorting algorithm, feature exist according to claim 4
In the calculation formula of the field base score value corresponding to the business scope is as follows:
Wherein,When being expressed as k-th of expert i-th of business scope being come jth position, corresponding to i-th of business scope
Field base score value;When representing that i-th of business scope is come jth position by k-th of expert, corresponding to i-th of business scope
Score value;M is expressed as the total number of business scope.
6. a kind of common people close friend's index assessment method based on big data sorting algorithm, feature exist according to claim 5
In the calculation formula of the importance value corresponding to the business scope is as follows:
Wherein, SiIt is expressed as the importance value corresponding to i-th of business scope;N is expressed as expert's total number of persons;bi(j)It is expressed as
When i-th of business scope comes jth position, the field base score value corresponding to i-th of business scope;NjIt represents to lead i-th of business
Domain comes expert's number of jth position.
7. according to a kind of any one of claim 1-6 common people close friend's index assessment methods based on big data sorting algorithm,
It is characterized in that, the importance value according to corresponding to work order to be determined and affiliated mood number of degrees, are calculated
The step for the first scoring rate corresponding to work order to be determined, it is specially:
After importance value corresponding to work order to be determined is multiplied with affiliated mood number of degrees, obtained product conduct
The first scoring rate corresponding to work order to be determined.
8. a kind of common people close friend's index assessment system based on big data sorting algorithm, which is characterized in that including:
Acquiring unit, for obtaining work order to be determined;
Extraction unit, for extracting the first business datum and the first emotion vocabulary number from the work order to be determined acquired
According to;
First processing units for the first business datum that basis extracts, judge the business scope belonging to work order to be determined,
Then the business scope according to belonging to work order to be determined and business scope score model, acquire corresponding to work order to be determined
Importance value;
Second processing unit, for according to the first business datum, the first emotion lexical data and the mood ranking score extracted
Class model judges the mood number of degrees belonging to work order to be determined;
Third processing unit, for the importance value according to corresponding to work order to be determined and affiliated mood number of degrees, meter
Calculate the first scoring rate obtained corresponding to work order to be determined;
Fourth processing unit carries out data for the first scoring rate corresponding to work order to be determined to be input in SVM classifier
Processing, so as to export common people close friend's index.
9. a kind of common people close friend's index assessment system based on big data sorting algorithm, feature exist according to claim 8
In further including the construction unit for building SVM classifier, the construction unit includes:
Acquisition module, for obtaining multiple sample work orders;
First processing module, for utilizing the second business datum and the second emotion vocabulary number extracted from multiple sample work orders
According to structure obtains mood grade separation model;
Second processing module, for using the importance value corresponding to multiple business scopes, structure to obtain business scope score
Model;
Third processing module for utilizing mood grade separation model and business scope score model, calculates multiple sample works
The second scoring rate corresponding to list;
Fourth processing module, for marking common people close friend's index corresponding to multiple sample work orders;
5th processing module, for by the second scoring rate corresponding to multiple sample work orders respectively constitute trained input data set and
Test input data set;
6th processing module, for by common people close friend's index composing training output data set corresponding to multiple sample work orders and survey
Try output data set;
7th processing module, for choosing a parameter to coming respectively as at the beginning of error penalty factor from preset multiple parameters centering
Initial value and nuclear parameter initial value;
8th processing module, for utilizing training input data set, training output data set, error penalty factor initial value and core
Initial parameter value is trained SVM classifier;
9th processing module, for using input data set and test output data set is tested, being obtained after training of judgement
Whether common people close friend's index that SVM classifier is exported meets preset accuracy rate requirement, if so, after currently training
SVM classifier of the obtained SVM classifier as required structure;It is re-executed corresponding to the 7th processing module conversely, then returning
Flow chart of data processing.
10. a kind of common people close friend's index assessment system based on big data sorting algorithm, which is characterized in that including:
At least one processor;
At least one processor, for storing at least one program;
When at least one program is performed by least one processor so that at least one processor is realized as weighed
Profit requires common people close friend's index assessment method based on big data sorting algorithm described in any one of 1-7.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711366824.0A CN108256552A (en) | 2017-12-18 | 2017-12-18 | Common people close friend's index assessment method and system based on big data sorting algorithm |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711366824.0A CN108256552A (en) | 2017-12-18 | 2017-12-18 | Common people close friend's index assessment method and system based on big data sorting algorithm |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108256552A true CN108256552A (en) | 2018-07-06 |
Family
ID=62723566
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201711366824.0A Pending CN108256552A (en) | 2017-12-18 | 2017-12-18 | Common people close friend's index assessment method and system based on big data sorting algorithm |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108256552A (en) |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2008082821A (en) * | 2006-09-27 | 2008-04-10 | Hitachi High-Technologies Corp | Defect classification method, its device, and defect inspection device |
CN103345525A (en) * | 2013-07-22 | 2013-10-09 | 苏州大学 | Method, device and processor for text categorization |
CN105760493A (en) * | 2016-02-18 | 2016-07-13 | 国网江苏省电力公司电力科学研究院 | Automatic work order classification method for electricity marketing service hot spot 95598 |
CN106529804A (en) * | 2016-11-09 | 2017-03-22 | 国网江苏省电力公司南京供电公司 | Client complaint early-warning monitoring analyzing method based on text mining technology |
CN106611375A (en) * | 2015-10-22 | 2017-05-03 | 北京大学 | Text analysis-based credit risk assessment method and apparatus |
CN106897792A (en) * | 2017-01-10 | 2017-06-27 | 广东广业开元科技有限公司 | A kind of structural fire protection risk class Forecasting Methodology and system |
-
2017
- 2017-12-18 CN CN201711366824.0A patent/CN108256552A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2008082821A (en) * | 2006-09-27 | 2008-04-10 | Hitachi High-Technologies Corp | Defect classification method, its device, and defect inspection device |
CN103345525A (en) * | 2013-07-22 | 2013-10-09 | 苏州大学 | Method, device and processor for text categorization |
CN106611375A (en) * | 2015-10-22 | 2017-05-03 | 北京大学 | Text analysis-based credit risk assessment method and apparatus |
CN105760493A (en) * | 2016-02-18 | 2016-07-13 | 国网江苏省电力公司电力科学研究院 | Automatic work order classification method for electricity marketing service hot spot 95598 |
CN106529804A (en) * | 2016-11-09 | 2017-03-22 | 国网江苏省电力公司南京供电公司 | Client complaint early-warning monitoring analyzing method based on text mining technology |
CN106897792A (en) * | 2017-01-10 | 2017-06-27 | 广东广业开元科技有限公司 | A kind of structural fire protection risk class Forecasting Methodology and system |
Non-Patent Citations (1)
Title |
---|
宋云波: "基于专家评分和回归分析的信息安全风险评估方法研究", 《中国优秀硕士学位论文全文数据库经济与管理科学辑》 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105184315B (en) | A kind of quality inspection processing method and system | |
CN105653444B (en) | Software defect fault recognition method and system based on internet daily record data | |
WO2017143919A1 (en) | Method and apparatus for establishing data identification model | |
CN108363717B (en) | Data security level identification and detection method and device | |
CN107403017A (en) | A kind of method that real-time news of intellectual analysis influences on financial market | |
CN106651057A (en) | Mobile terminal user age prediction method based on installation package sequence table | |
CN107203467A (en) | The reference test method and device of supervised learning algorithm under a kind of distributed environment | |
CN112700325A (en) | Method for predicting online credit return customers based on Stacking ensemble learning | |
CN107833059B (en) | Service quality evaluation method and system for customer service | |
CN107122432A (en) | CSR analysis method, device and system | |
CN107944761A (en) | Early warning and monitoring analysis method is complained based on artificial intelligence protection of consumers' rights index enterprise | |
WO2024067387A1 (en) | User portrait generation method based on characteristic variable scoring, device, vehicle, and storage medium | |
CN107992613A (en) | A kind of Text Mining Technology protection of consumers' rights index analysis method based on machine learning | |
CN115396389A (en) | Internet of things information technology customer service system based on emotion energy perception | |
CN108509588B (en) | Lawyer evaluation method and recommendation method based on big data | |
CN107766560A (en) | The evaluation method and system of customer service flow | |
CN112116168B (en) | User behavior prediction method and device and electronic equipment | |
CN113704389A (en) | Data evaluation method and device, computer equipment and storage medium | |
WO2024055603A1 (en) | Method and apparatus for identifying text from minor | |
CN107886233B (en) | Service quality evaluation method and system for customer service | |
CN110096708A (en) | A kind of determining method and device of calibration collection | |
CN108256552A (en) | Common people close friend's index assessment method and system based on big data sorting algorithm | |
CN109886288A (en) | A kind of method for evaluating state and device for power transformer | |
CN114443930A (en) | News public opinion intelligent monitoring and analyzing method, system and computer storage medium | |
CN113887994A (en) | Failure mode risk assessment method and system based on Internet comment mining |
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: 20180706 |
|
RJ01 | Rejection of invention patent application after publication |