The content of the invention
Recommend method and device it is an object of the invention to provide a kind of interpreter, the problem of to solve above-mentioned.
A kind of interpreter is provided in an embodiment of the present invention recommends method, including:
The deviation of interpreter's data item occurrence to be matched and standard interpreter's data item occurrence is obtained, and according to the deviation iteration meter of acquisition
Calculate the matching degree of interpreter's data item occurrence to be matched and standard interpreter's data item occurrence;Wherein, interpreter's data item occurrence to be matched includes turning over
Translate quality, translation unit price and translation speed;Standard interpreter data item occurrence includes standard translation quality, standard translation price and standard
Translation speed;Interpreter to be matched is determined according to the matching degree of iterative calculation.
Further, the deviation of interpreter's data item occurrence to be matched and standard interpreter's data item occurrence is obtained, and according to the inclined of acquisition
Difference iterative calculation interpreter's data item occurrence to be matched before the matching degree of standard interpreter's data item occurrence with also including:To interpreter to be matched
Data item occurrence and standard interpreter's data item occurrence are normalized.
Further, interpreter's data item occurrence to be matched and standard interpreter's data item occurrence are normalized including:
Data item occurrence is mapped to nor_low to nor_high from the minimum value min of preset range between maximum max
Between, mapping equation is:
Wherein, nor_low and nor_high are respectively the upper and lower interval after normalization, and x and y are variable, and x is normalizing
Value before change, interval range is min-max, and y is the value after normalization, and interval range is nor_low-nor_high, formula (1-1)
It is positive condition for occurrence, it is negative sense condition that formula (1-2), which is used for occurrence, and min-max default span is data
The number range of entry value.
Further, the deviation of interpreter's data item occurrence to be matched and standard interpreter's data item occurrence is obtained, and according to the inclined of acquisition
The matching degree of difference iterative calculation interpreter's data item occurrence to be matched and standard interpreter's data item occurrence includes:
Bias div_left and lower bias div_right in calculating, calculation formula is:
Div_left=cnor_i-nor_low;Div_right=nor_high-cnor_i;
In formula, cnor_iFor the matched value after normalization;
By matched value mnor_iBe converted to standard matched value m 'nor_i, conversion formula is:
Matching amplification coefficient amp, calculation formula are determined according to upper bias div_left and lower bias div_right
For:
According to the standard matched value m ' of acquisitionnor_iAnd matching amplification coefficient amp, ask for matching using cycloid Equation Iterative
Spend matiApproximate solution, wherein, the accuracy acc of iterative is set smaller than or equal to 10-3, π is pi, i ∈ 1,
2 ..., k }, k is that ability item number is natural number.
Further, determine that interpreter to be matched includes according to the matching degree of iterative calculation:
According to matching degree matiAsk for matching score value mscore, matching score value mscore calculation formula is:
In formula, { w1,w2,…,wkFor the intrinsic weight of interpreter each ability, k is that ability item number is natural number, mati
For matching degree;
Height order according to the matching score value mscore asked for selects interpreter to be matched.
Embodiments of the invention additionally provide a kind of interpreter's recommendation apparatus, including:
Matching degree computing module, the deviation for obtaining interpreter's data item occurrence to be matched and standard interpreter's data item occurrence, and
The matching degree of interpreter's data item occurrence to be matched and standard interpreter's data item occurrence is iterated to calculate according to the deviation of acquisition;Wherein, treat
Include translation quality, translation unit price and translation speed with interpreter's data item occurrence;Standard interpreter data item occurrence includes standard translation matter
Amount, standard translation price and standard translation speed;
Interpreter's determining module to be matched, for determining interpreter to be matched according to the matching degree of iterative calculation.
Further, the device also includes:Normalized module, for interpreter's data item occurrence to be matched and standard interpreter
Data item occurrence is normalized.
Further, normalized module specifically for:
Data item occurrence is mapped to nor_low to nor_high from the minimum value min of preset range between maximum max
Between, mapping equation is:
Wherein, nor_low and nor_high are respectively the upper and lower interval after normalization, and x and y are variable, and x is normalizing
Value before change, interval range is min-max, and y is the value after normalization, and interval range is nor_low-nor_high, formula (1-1)
It is positive condition for occurrence, it is negative sense condition that formula (1-2), which is used for occurrence, and min-max default span is data
The number range of entry value.
Further, matching degree computing module specifically for:
Bias div_left and lower bias div_right in calculating, calculation formula is:
Div_left=cnor_i-nor_low;Div_right=nor_high-cnor_i;
In formula, cnor_iFor the matched value after normalization;
By matched value mnor_iBe converted to standard matched value m 'nor_i, conversion formula is:
Matching amplification coefficient amp, calculation formula are determined according to upper bias div_left and lower bias div_right
For:
According to the standard matched value m ' of acquisitionnor_iAnd matching amplification coefficient amp, ask for matching using cycloid Equation Iterative
Spend matiApproximate solution, wherein, the accuracy acc of iterative is set smaller than or equal to 10-3, π is pi, i ∈ 1,
2 ..., k }, k is that ability item number is natural number.
Further, interpreter's determining module to be matched specifically for:
According to matching degree matiAsk for matching score value mscore, matching score value mscore calculation formula is:
In formula, { w1,w2,…,wkFor the intrinsic weight of interpreter each ability, k is that ability item number is natural number, mati
For matching degree;
Height order according to the matching score value mscore asked for selects interpreter to be matched.
Interpreter provided in an embodiment of the present invention recommends method and device compared with prior art, by using interpreter to be matched
Matching degree that the deviation iteration of data item occurrence and standard interpreter's data item occurrence is asked for extracts interpreter to be matched, reduces to be matched
The mistake recovery rate of interpreter, it is ensured that the reasonable distribution of resource and utilization.
Embodiment
In the following detailed description, a large amount of specific details are proposed, in order to provide thorough understanding of the present invention.But,
It will be understood by those within the art that can also implement the present invention even if without these specific details.In other cases, without detailed
Well-known method, process, component and circuit are carefully described, in order to avoid influence the understanding of the present invention.
In order to be better understood from the illustrative embodiment in the present invention, below to some in illustrative embodiment of the present invention
Term is briefly described.
The present invention is described in further detail below by specific embodiment and with reference to accompanying drawing.
Join shown in Fig. 1, Fig. 1 shows that interpreter of the present invention recommends the flow chart of method first embodiment.
Present embodiments provide a kind of interpreter and recommend method, including:
Step S102, obtains the deviation of interpreter's data item occurrence to be matched and standard interpreter's data item occurrence, and according to acquisition
Deviation iterates to calculate the matching degree of interpreter's data item occurrence to be matched and standard interpreter's data item occurrence;Wherein, interpreter's data to be matched
Entry value includes translation quality, translation unit price and translation speed;Standard interpreter data item occurrence includes standard translation quality, standard translation
Price and standard translation speed;Interpreter to be matched is determined according to the matching degree of iterative calculation;
Step S104, interpreter to be matched is determined according to the matching degree of iterative calculation.
The present embodiment is asked for by using the deviation iteration of interpreter's data item occurrence to be matched and standard interpreter's data item occurrence
Matching degree extracts interpreter to be matched, reduces the mistake recovery rate of interpreter to be matched, it is ensured that the reasonable distribution of resource and utilization.
Join shown in Fig. 2, Fig. 2 shows that interpreter of the present invention recommends the flow chart of method second embodiment.
In the present embodiment, possess comparable between interpreter's data item occurrence to be matched and standard interpreter's data item occurrence in order that obtaining
Property, it can be uniformly processed in identical quantity, obtain interpreter's data item occurrence to be matched and standard interpreter's data item occurrence
Deviation and according to the deviation of acquisition iterate to calculate interpreter's data item occurrence to be matched with before the matching degree of standard interpreter's data item occurrence
It can also include:
Step S101, interpreter's data item occurrence to be matched and standard interpreter's data item occurrence are normalized.
The detailed process of above-mentioned normalized includes:
Data item occurrence is mapped to nor_low to nor_high from the minimum value min of preset range between maximum max
Between, mapping equation is:
Wherein, nor_low and nor_high are respectively the upper and lower interval after normalization, and x and y are variable, and x is normalizing
Value before change, interval range is min-max, and y is the value after normalization, and interval range is nor_low-nor_high, formula (1-1)
It is positive condition for occurrence, it is negative sense condition that formula (1-2), which is used for occurrence, and min-max default span is data
The number range of entry value.
In the present embodiment, the deviation of interpreter's data item occurrence to be matched and standard interpreter's data item occurrence is obtained, and according to obtaining
The matching degree that the deviation taken iterates to calculate interpreter's data item occurrence to be matched and standard interpreter's data item occurrence is specifically included:
Bias div_left and lower bias div_right in calculating, calculation formula is:
Div_left=cnor_i-nor_low;Div_right=nor_high-cnor_i;
In formula, cnor_iFor the matched value after normalization;
By matched value mnor_iBe converted to standard matched value m 'nor_i, conversion formula is:
Matching amplification coefficient amp, calculation formula are determined according to upper bias div_left and lower bias div_right
For:
According to the standard matched value m ' of acquisitionnor_iAnd matching amplification coefficient amp, ask for matching using cycloid Equation Iterative
Spend matiApproximate solution, wherein, the accuracy acc of iterative is set smaller than or equal to 10-3, π is pi, i ∈ 1,
2 ..., k }, k is that ability item number is natural number.Specific iterative algorithm is as follows:
Begin
Acc is set smaller than equal to 10-3Value;
X_low=0;X_high=2 π;X=(x_low+x_high)/2;
while(|m′nor_i- (x-sinx) | > acc)
if(m′nor_i> (x-sinx))
X_low=(x_low+x_high)/2;X=(x_low+x_high)/2;}
else{
X_high=(x_low+x_high)/2;X=(x_low+x_high)/2;}
}
mati=(amp* (1-cosx)/2+1)/(amp+1);
End
Wherein, acc is accuracy, and x is the value in data item occurrence preset range, and x_low is in data item occurrence preset range
Minimum value, x_high be data item occurrence preset range in maximum, mati is matching degree.
The present embodiment solves standard interpreter data item occurrence and interpreter's data to be matched using the characteristic of cycloid (cycloid)
The matching problem of entry value.
The parametric equation of cycloid is:
Wherein r is the radius of circle, radius rollings one week and the angle of vertical curve of the θ for circle.Cycloid has a key property, its
During y values are gradually decreased from peak 2r to minimum point 0, its rate of descent is more and more faster.Using this characteristic, it will mark
Each data item occurrence of quasi- interpreter is set in peak, and its matching degree is set to 1, and interpreter's data item occurrence to be matched is with being matched data
Entry value difference is smaller, and its matching degree is closer to 1, and the bigger matching degree of difference is smaller, and diminishing for this matching degree is not at the uniform velocity
Reduce, but as the increase matching degree of difference can accelerate to reduce, this ensures that more meeting standard interpreter's data item occurrence
The probability that interpreter to be matched is selected is bigger.
In the present embodiment, determine that matching interpreter specifically includes according to the choosing of the matching degree of iterative calculation:
According to matching degree matiAsk for matching score value mscore, matching score value mscore calculation formula is:
In formula, { w1,w2,…,wkFor the intrinsic weight of interpreter each ability, k is that ability item number is natural number, mati
For matching degree;
Height order according to the matching score value mscore asked for selects interpreter to be matched.
Join shown in Fig. 3, Fig. 3 shows the structured flowchart of interpreter's recommendation apparatus first embodiment of the present invention.
Embodiments of the invention additionally provide a kind of interpreter's recommendation apparatus, including:
Matching degree computing module 32, the deviation for obtaining interpreter's data item occurrence to be matched and standard interpreter's data item occurrence,
And the matching degree of interpreter's data item occurrence to be matched and standard interpreter's data item occurrence is iterated to calculate according to the deviation of acquisition;Wherein, treat
Matching interpreter's data item occurrence includes translation quality, translation unit price and translation speed;Standard interpreter data item occurrence includes standard translation
Quality, standard translation price and standard translation speed;
Object Selection module 34 to be matched, for determining interpreter to be matched according to the matching degree of iterative calculation.
The present embodiment utilizes interpreter's data item occurrence to be matched and standard interpreter's data item occurrence by matching degree computing module 32
The matching degree asked for of deviation iteration, by the selective extraction of interpreter's determining module 34 to be matched interpreter to be matched, reduce and treat
Mistake recovery rate with interpreter, it is ensured that the reasonable distribution of resource and utilization.
Join shown in Fig. 4, Fig. 4 shows the structured flowchart of interpreter's recommendation apparatus second embodiment of the present invention.
In the present embodiment, in order that obtaining interpreter's data item occurrence to be matched and standard interpreter's data item occurrence to be matched
Between possess comparativity, can be uniformly processed in identical quantity, the device can also include:Normalized module
31, for interpreter's data item occurrence to be matched and Standard object data entry value to be normalized.
The normalized module 31 specifically for:
Data item occurrence is mapped to nor_low to nor_high from the minimum value min of preset range between maximum max
Between, mapping equation is:
Wherein, nor_low and nor_high are respectively the upper and lower interval after normalization, and x and y are variable, and x is normalizing
Value before change, interval range is min-max, and y is the value after normalization, and interval range is nor_low-nor_high, formula (1-1)
It is positive condition for occurrence, it is negative sense condition that formula (1-2), which is used for occurrence, and min-max default span is data
The number range of entry value.
In the present embodiment, the matching degree computing module 32 specifically for:
Bias div_left and lower bias div_right in calculating, calculation formula is:
Div_left=cnor_i-nor_low;Div_right=nor_high-cnor_i;
In formula, cnor_iFor the matched value after normalization;
By matched value mnor_iBe converted to standard matched value m 'nor_i, conversion formula is:
Matching amplification coefficient amp, calculation formula are determined according to upper bias div_left and lower bias div_right
For:
According to the standard matched value m ' of acquisitionnor_iAnd matching amplification coefficient amp, ask for matching using cycloid Equation Iterative
Spend matiApproximate solution, wherein, the accuracy acc of iterative is set smaller than or equal to 10-3, π is pi, i ∈ 1,
2 ..., k }, k is that ability item number is natural number.Specific iterative algorithm is as follows:
Begin
Acc is set smaller than equal to 10-3Value;
X_low=0;X_high=2 π;X=(x_low+x_high)/2;
while(|m′nor_i- (x-sinx) | > acc)
if(m′nor_i> (x-sinx))
X_low=(x_low+x_high)/2;X=(x_low+x_high)/2;}
else{
X_high=(x_low+x_high)/2;X=(x_low+x_high)/2;}
}
mati=(amp* (1-cosx)/2+1)/(amp+1);
End
Wherein, acc is accuracy, and x is the value in data item occurrence preset range, and x_low is in data item occurrence preset range
Minimum value, x_high be data item occurrence preset range in maximum, mati is matching degree.
In the present embodiment, interpreter's determining module 34 to be matched specifically for:
According to matching degree matiAsk for matching score value mscore, matching score value mscore calculation formula is:
In formula, { w1,w2,…,wkFor the intrinsic weight of interpreter each ability, k is that ability item number is natural number, mati
For matching degree;
Height order according to the matching score value mscore asked for selects interpreter to be matched.
In order to more intuitively describe the method that the present invention is provided, it is described further below with specific data.
(1) input data is handled
1st, standard interpreter data are referring to table 1
The standard interpreter's data of table 1
Standard interpreter |
Quality requirement |
Rate request |
Price request |
Doc1 |
85 |
4000 |
110 |
2nd, interpreter's data to be matched are referring to table 2
The interpreter's data to be matched of table 2
Object to be matched |
Quality |
Speed |
Price |
IP1 |
90 |
5100 |
110 |
IP2 |
95 |
5500 |
120 |
IP3 |
85 |
4200 |
103 |
IP4 |
88 |
4400 |
109 |
IP4 |
99 |
6000 |
120 |
3rd, normalized
According to following formula:
Each item data of interpreter's data is mapped between nor_low-nor_high between min-max so that these
Matching item data possesses comparativity, can be handled in identical quantity.
Nor_low=50, nor_high=100 are taken, quality, speed, price this 3 are normalized:
1) quality score is the weighted average for the quality score that contribution is translated according to interpreter's history, is one in 50-100
Data, without processing.
2) translation speed is positive condition, and with formula (1-1), preset range is 1000-6000, and mapping equation is:Y=x/
100+40。
3) translation price is negative sense condition, and with formula (1-2), preset range is 80-130, and mapping equation is:Y=180-x.
After normalized, referring to table 3.
Interpreter's data to be matched after the normalized of table 3
Object to be matched |
Quality |
Speed |
Price |
IP1 |
90 |
90 |
5100 |
91 |
110 |
70 |
IP2 |
95 |
95 |
5500 |
95 |
118 |
62 |
IP3 |
85 |
85 |
4200 |
82 |
103 |
77 |
IP4 |
88 |
88 |
4400 |
84 |
109 |
71 |
IP5 |
99 |
99 |
5900 |
99 |
120 |
60 |
(2) interpreter to be matched is determined by matching degree
Weight setting:Quality:0.5;Speed:0.2;Price:0.3.
Because IP2 and IP5 variation in prices closes imposing a condition less than 110, IP2 and IP2 are excluded.
According to the data of interpreter to be matched, according to foregoing iterative algorithm, interpreter's matching degree of each to be matched is calculated, is obtained
Degrees of data is matched to interpreter to be matched, 4 are shown in Table.
The interpreter to be matched of table 4 matches degrees of data
According to the weight of setting, interpreter's score to be matched is calculated according to foregoing matching score value mscore calculation formula:
IP1 scores:90*0.962*0.5+91*0.77*0.2+70*0.3=78.304
IP3 scores:85*0.5+82*0.995*0.2+77*0.963*0.3=81.063
IP4 scores:88*0.987*0.5+84*0.969*0.2+71*0.998=80.965
According to score, according to the height (preparatory condition) of matching score value determine the order of object to be matched for IP3 → IP4 →
IP1。
(3) object to be matched is determined by intrinsic weight
IP1 scores:90*0.5+91*0.2+70*0.3=84.2
IP2 scores:95*0.5+95*0.2+62*0.3=85.1
IP3 scores:85*0.5+82*0.2+77*0.3=82
IP4 scores:88*0.5+84*0.2+71*0.3=82.1
IP5 scores:99*0.5+99*0.2+60*0.3=87.3
Because IP2 and IP5 variation in prices closes the requirement less than 110, IP2 and IP5 are excluded, is successively arranged according to obtaining
, the order of preference of interpreter so to be matched is:IP1→IP4→IP3.
By (three) it can be seen that, although interpreter IP1 every capacity index is all very high, by the method for intrinsic weight first
Can selecting object IP1, but the matching degree ratio IP3 and IP4 of each data item of itself and standard interpreter is small, and according to of the invention real
Determining that the method for interpreter to be matched can be chosen by matching degree and more conforming to the interpreter to be matched of standard interpreter for example offer is provided,
Improve matching accuracy rate, it is to avoid the wasting of resources.
The preferred embodiments of the present invention are the foregoing is only, are not intended to limit the invention, for the skill of this area
For art personnel, the present invention can have various modifications and variations.Within the spirit and principles of the invention, that is made any repaiies
Change, equivalent substitution, improvement etc., should be included in the scope of the protection.