CN102819605A - Adaptability matching method - Google Patents

Abstract

The invention provides an adaptability matching method, which comprises the following steps of: selecting a group of parameters for each object in at least two objects, and setting the group of parameters as a group of parameter vectors of the object; setting a weight for each of the group of parameter vectors; multiplying each of the group of parameter vectors and the weights thereof so as to obtain a corresponding group of weighted parameter vectors; and calculating an included angle between the weighted parameter vectors and a target vector, wherein the size of the included angle refers to the similarity between the object and the target. According to the adaptability matching method, a brand new technical development direction is provided. The related parameters of an entity object are subjected to vectorization, so that the matching degree of each of a group of entity objects relative to a certain target object is effectively measured. The technology can be applied to the application fields of searching, electronic commerce, electronic procurement system and the like, and the data separation speed and performance are improved.

Description

The adaptability matching process

Technical field

The present invention relates to a kind of adaptability matching process, relate in particular to a kind of adaptability matching process that is applicable to Object Selection.

Background technology

Along with the development of infotech, the electronization of product has become main flow trend.For example, in the traditional business model of the just deep change of technology such as numeral search, it is not only with whole data handling procedure robotization, transparence; Increase the benefit, reduce cost, what is more important makes enterprise rely on e-commerce platform; The focus of work turns to the source of seeking and tactical management from execution; Thereby continue to optimize supply, improve concentrated management and control ability, make competitive flexible supply chain business.

For example, in digital search field, how confirm it is a very important technology to what object carried out matching degree or similarity.For example, traditional traditional adaptability matching process such as matching Uniqueness, gray scale similarity that have are weighed the similarity between an entity object and the target, to help to improve the accuracy of the output of searching for or retrieving.But, along with the deep development of digital search technique, adaptability matching technique more, that upgrade is also progressively appearring.

Summary of the invention

For example, the invention provides a kind of brand-new adaptability matching technique, this technology is based on the measurement that the vectorization of being carried out parameter by object is relatively realized matching degree.

Particularly, the invention provides a kind of adaptability matching process, may further comprise the steps:

Being the one group of parameter of each Object Selection at least two objects, is one group of parameter vector of this object with this group parameter setting;

Be the weights of each setting in said one group of parameter vector;

Each and weights thereof in said one group of parameter vector are multiplied each other, to obtain corresponding one group of weighting parameters vector; And

Calculate the angle between the vector of said weighting parameters vector and target, the size of said angle representes that said object is with the similarity degree between the said target.

According to a preferred embodiment of the present invention, in above-mentioned adaptability matching process, said angle is more little, and said similarity is high more.

According to a preferred embodiment of the present invention, in above-mentioned adaptability matching process, the span of said angle is 0 to 1.

According to a preferred embodiment of the present invention; In above-mentioned adaptability matching process, the weights of each in said one group of parameter vector are to confirm according to the sum of the object under selecteed number of times, these weights in said at least two objects of the pairing parameter of these weights and the sum that comprises the object of the pairing parameter of these weights.

According to a preferred embodiment of the present invention; In above-mentioned adaptability matching process; Angle between the vector that calculates said weighting parameters vector and target; The size of said angle representes that this method also comprises after the step of said object with the similarity degree between the said target: export the tabulation that said angle is less than or equal to all objects of a threshold value, said tabulation comprises the numerical value of the angle of these objects.

Adaptability matching process of the present invention has proposed a kind of brand-new technology developing direction.Carry out vectorization through correlation parameter and come each the matching degree in a collection of entity object of valid metric with respect to a certain destination object to entity object.This technology can be applicable in search, ecommerce, e-procurement systems or the like the application, improves the speed and the performance of data separating.

Should be appreciated that the above generality of the present invention is described and following detailed description all is exemplary and illustrative, and be intended to further explanation is provided for as claimed in claim the present invention.

Description of drawings

Accompanying drawing mainly is to be used to provide the present invention is further understood.Accompanying drawing shows embodiments of the invention, and plays the effect of explaining the principle of the invention with this instructions.In the accompanying drawing:

Fig. 1 schematically shows the process flow diagram according to the basic step of adaptability matching process of the present invention.

Embodiment

Describe technical scheme of the present invention in detail below in conjunction with accompanying drawing.

The invention provides a kind of brand-new adaptability matching technique.This technology is based on the measurement that the vectorization of being carried out parameter by object is relatively realized matching degree.Fig. 1 schematically shows the process flow diagram according to the basic step of adaptability matching process of the present invention.As shown in the figure, adaptability matching process 100 of the present invention mainly may further comprise the steps:

Step 101: being the one group of parameter of each Object Selection at least two objects, is one group of parameter vector of this object with this group parameter setting;

Step 102: be the weights of each setting in said one group of parameter vector;

Step 103: each and weights thereof in said one group of parameter vector are multiplied each other, to obtain corresponding one group of weighting parameters vector;

Step 104: calculate the angle between the vector of said weighting parameters vector and target, the size of said angle representes that said object is with the similarity degree between the said target; And

Step 105: export the tabulation that said angle is less than or equal to all objects of a threshold value, said tabulation comprises the numerical value of the angle of these objects.

Particularly, for example, in step 101, can regard as for object and be the multi-C vector space, (Vector Space Model, VSM) in VSM, it is by one group of vector (T that object is seen as to make up vector space model ₁, T ₂... T _n) constitute.For each parameter vector Ti, all compose with certain weight W according to its significance level _iCan regard it as n dimension coordinate is W ₁, W ₂... W _nTherefore, each object all can be mapped as a point in the vector space of being opened by one group of vector.For all objects available feature vector (T all ₁W ₁, T ₂W ₂..T _nW _n) expression.Such as, above-mentioned object can be the notebook product, the user is one group of parameter of its selection, can comprise the technical parameter that is easy to quantize that its processor host frequency, hard-disk capacity, memory size or the like are main.

The target of supposing the user is D, and proper vector is M, and the angle between both similarity degree availability vectors is measured, and the bright similarity of novel is high more more for angle, and calculating formula of similarity is following:

$\mathrm{Sim}(D,M)=\mathrm{COS}(D,M)=\underset{k=1}{\overset{n}{\mathrm{\Σ}}}{W}_{\mathrm{dk}}\×W_{\mathrm{mk}}/\sqrt{\underset{k=1}{\overset{n}{\mathrm{\Σ}}}{W}_{\mathrm{dk}}^2}\×\sqrt{\underset{k=1}{\overset{n}{\mathrm{\Σ}}}{W}_{\mathrm{mk}}^2}$

The span of this angle is 0 to 1.When angle is zero, Sim (D _MAND, M _ODEL)=1 representes that both similarities are in full accord.When angle increases gradually, Sim (D _MAND, M _ODEL) reduce gradually, trend zero representes that both similarities are very low.

In addition, in step 102, according to a preferred embodiment, can be with each the weights of setting according to following equality in said one group of parameter vector:

$W_{\mathrm{ik}}=f_{\mathrm{ik}}\log [\frac{N}{n_k}+0.01]/\sqrt{\underset{k=1}{\overset{n}{\mathrm{\Σ}}}{\left(f_{\mathrm{ki}}\right)}^2{\log }^2[\frac{N}{n_k}+0.01]}$

In the formula: f _IkThe selecteed number of times in said at least two objects of representing the pairing parameter of these weights; N representes the sum of the object under these weights; Nk representes to comprise the sum of the object of the pairing parameter of these weights.

Adaptability matching process of the present invention has proposed a kind of brand-new technology developing direction.Carry out vectorization through correlation parameter and come each the matching degree in a collection of entity object of valid metric with respect to a certain destination object to entity object.This technology can be applicable in search, ecommerce, e-procurement systems or the like the application, improves the speed and the performance of data separating.

The foregoing description provides to those of ordinary skills and realizes or use of the present invention; Those of ordinary skills can be under the situation that does not break away from invention thought of the present invention; The foregoing description is made various modifications or variation; Thereby protection scope of the present invention do not limit by the foregoing description, and should be the maximum magnitude that meets the inventive features that claims mention.

Claims (5)

1. an adaptability matching process is characterized in that, may further comprise the steps:

Being the one group of parameter of each Object Selection at least two objects, is one group of parameter vector of this object with this group parameter setting;

Be the weights of each setting in said one group of parameter vector;

Each and weights thereof in said one group of parameter vector are multiplied each other, to obtain corresponding one group of weighting parameters vector; And

Calculate the angle between the vector of said weighting parameters vector and target, the size of said angle representes that said object is with the similarity degree between the said target.

2. adaptability matching process as claimed in claim 1 is characterized in that said angle is more little, and said similarity is high more.

3. adaptability matching process as claimed in claim 1 is characterized in that, the span of said angle is 0 to 1.

4. adaptability matching process as claimed in claim 1; It is characterized in that the weights of each in said one group of parameter vector are to confirm according to the sum of the object under selecteed number of times, these weights in said at least two objects of the pairing parameter of these weights and the sum that comprises the object of the pairing parameter of these weights.

5. adaptability matching process as claimed in claim 1; It is characterized in that; Angle between the vector that calculates said weighting parameters vector and target, the size of said angle representes that this method also comprises after the step of said object with the similarity degree between the said target:

Export said angle and be less than or equal to the tabulation of all objects of a threshold value, said tabulation comprises the numerical value of the angle of these objects.

Images