CN103279535A - Method for recommending potential partners for patentee - Google Patents

Abstract

The invention relates to a method for recommending potential partners for a patentee. The method includes firstly judging whether relative number ranking of patents of the patentee under each small class level IPC (information procession center) exceeds a set threshold value or not and a field represented by the IPC is attributed to a core research field or non-core research field of the patentee. In the core research field, other patentees with complementary skills are recommended to the patentee; and in the non-core research field, other patentees sharing similar technology with the existing partners are recommended to the patentee. The method for recommending the potential partners for the patentee proposed for the first time solves the problem of complementary technology degree calculation of the other patentees and the given patentee in the core research field and the problem of predicted values of cooperative relationship between the other patentees and the given patentee by taking the existing partners as media in the non-core research field. The method for recommending the potential partners for the patentee has the advantages of good effect in recommending the potential partners for the patentee.

Description

The potential affiliate's recommend method of a kind of patentee

Technical field

The present invention relates to the potential affiliate's of a kind of patentee method, especially relate to the potential affiliate's recommend method of a kind of patentee.

Background technology

Since World War II finished, various countries greatly developed economy and science and technology, and economy and science and technology are moved towards to merge gradually, and scientific and technological economic implication is increasing, and the global economy activity changes to kownledge economy from material economy.The development of society also enters a fast-changing era of knowledge-driven economy thereupon.

Meanwhile, information management also becomes for expanding economy and becomes more and more important.Current kownledge economy fast development, the source of enterprise core competence generally all is technological innovation, an enterprise only constantly brings forth new ideas and just can not eliminated by market.The enterprise that some are in the relation of vying each other originally cooperates aspect intellecture property one after another, and the intellecture property cooperation becomes a kind of important techniques innovation organizational form, more and more is familiar with by each side and payes attention to.

It is increasing that affiliate's demand is looked for by enterprise, and in a period of time from now on, demand still has the space of rising at least.

Enterprise relates in particular to technology aspect competitive intelligence competitive intelligence aspect most important information is patent, so patent analyses becomes the part of information analysis in the competitive intelligence naturally.Patent information is analyzed from proposing to develop so far, and it is analyzed content and mainly comprises three aspects: the discovery potential rival reaches to be analyzed, finds potential affiliate, the analysis of industry patent statute and monitoring industry are changed the rival.From the patent text analysis, advance series of analysis and handled analysis again, finally can comprise that the existing affiliate of rival analyzes and potential affiliate finds for enterprise obtains competitive intelligence.

Use the automatic analysis of computer software not only can save the efficient that manpower improves patent analyses greatly, can also in the magnanimity patent data, excavate some association knowledge that imply, with close friend's interface, mode represents analysis result intuitively.

Summary of the invention

The present invention solves the existing in prior technology technical matters; Can continue the hypothesis of research in the prior art field based on the patentee, provide a kind of each research field with the patentee to be divided into core research field or non-core research field, gathered the method for recommending potential affiliate from its core research field set and non-core research field respectively.

Above-mentioned technical matters of the present invention is mainly solved by following technical proposals:

The potential affiliate's recommend method of a kind of patentee is characterized in that, based on definition: patent file set D={d ₁, d ₂..., d _n, the patentee gathers P={P ₁, P ₂..., P _m; Wherein, d _iThe document content of expression patent i; P _jRepresent j patentee; According to the relative patent quantity rank of given patentee under each IPC classification number that its patent relates to, the field of IPC classification number representative is classified as given patentee's core research field or non-core research field, the patentee of core research field recommended technology complementation, recommend the patentee similar to its existing affiliate's technology in non-core research field; Comprise and judge that research field is the step in core realm or non-core field, core research field given patentee is the step that given patentee recommends potential affiliate: and be the step that given patentee recommends potential affiliate in given patentee's non-core research field, specifically comprise:

Judge that research field is the step of core research field or non-core research field:

Step 1.1 is analyzed other IPC of group level under the patent of given patentee q, obtains the IPC S set ₁

Step 1.2, pair set S ₁In each IPC, the patent quantity of adding up all patentees under this IPC, by the descending sort of patent quantity to patentee's rank;

Step 1.3, calculate q at the relative patent quantity rank R (q) in this IPC representative field according to the ranking result of step 1.2, if R (q) surpasses setting threshold, then this field is the core research field of q, if R (q) does not surpass setting threshold, then this field is the non-core research field of q;

Step 1.4, pair set S ₁In after each IPC analyzes, obtain the core research field S set of q ₂With non-core research field S set ₃

Core research field given patentee is the step that given patentee recommends potential affiliate:

Step 2.1 is analyzed S set ₂In patentee under each IPC, if patentee one hurdle of the patent file of q not only exists q to also have other patentees, then this patentee is considered as having cooperative relationship with q, the existing affiliate of core research field who obtains q gathers P ₁, screening patent quantity is no more than other patentees of setting threshold, and remaining patentee constitutes core research field candidate and gathers C ₁

Step 2.2, pair set C ₁In each patentee v, collect it and q in S set ₂In patent file under each IPC, with the document vectorization, calculate the document similarity, according to similarity with clustering documents;

Step 2.3 according to the cluster result of step 2.2, is calculated the technology complementation degree of q and v;

Step 2.4 according to the result of calculation of step 2.3, is taken all factors into consideration v in S set ₂Under research and development strength, inventor's quantity, affiliate's quantity, calculate the possibility that v becomes the q affiliate;

Step 2.5, to the result of calculation of step 2.4, descending sort is got preceding K patentee as the potential affiliate of core research field of q;

Non-core research field given patentee is the step that given patentee recommends potential affiliate:

Step 3.1 is analyzed S set ₃In patentee under each IPC, if patentee one hurdle of the patent file of q not only exists q to also have other patentees, then this patentee is considered as having cooperative relationship with q, the existing affiliate of non-core research field who obtains q gathers P ₂, other patentees constitute non-core research field candidate and gather C ₂

Step 3.2, pair set P ₂In each patentee k, calculate the cooperative relationship intensity of k and q;

Step 3.3, pair set C ₂In each patentee c, calculate he with the set P ₂In each patentee's technology similarity;

Step 3.4 is according to the cooperative relationship intensity of step 3.2 calculating and the c and set P of step 3.3 calculating ₂In each patentee's technology similarity, calculate the cooperative relationship prediction of strength value of c and q;

Step 3.5 according to the result of calculation of step 3.4, is taken all factors into consideration c in S set ₃Under research and development strength, inventor's quantity, affiliate's quantity, calculate the possibility that c becomes the q affiliate;

Step 3.6, to the result of calculation of step 3.5, descending sort is got preceding K patentee as the potential affiliate in non-core field of q.

The potential affiliate's recommend method of above-mentioned a kind of patentee, described step 1.1 concrete operation method is:

Analyze affiliated other IPC of group level of patent of q, obtain the IPC S set ₁

The potential affiliate's recommend method of above-mentioned a kind of patentee, described step 1.2 concrete operation method is:

Pair set S ₁In each IPC, the patent quantity of adding up all patentees under this IPC, by the descending sort of patent quantity to patentee's rank.

The potential affiliate's recommend method of above-mentioned a kind of patentee, described step 1.3 concrete operation method is:

Calculate q at the relative patent quantity rank R (q) in this IPC representative field according to the ranking result of step 1.2, its calculating formula is:

$R\left(q\right)=\frac{\mathrm{Rank}\left(q\right)}{N}$ Formula one;

Wherein, Rank (q) is the rank of q, and N is all the patentee's quantity under this IPC;

If R (q) surpasses setting threshold, then this field is the core research field of q, if R (q) does not surpass setting threshold, then this field is the non-core research field of q.

The potential affiliate's recommend method of above-mentioned a kind of patentee, described step 1.4 concrete operation method is:

Pair set S ₁In after each IPC analyzes, obtain the core research field S set of q ₂With non-core research field S set ₃

The potential affiliate's recommend method of above-mentioned a kind of patentee, described step 2.1 concrete operation method is:

Analyze S set ₂In patentee under each IPC, if patentee one hurdle of the patent file of q not only exists q to also have other patentees, then this patentee is considered as having cooperative relationship with q, the existing affiliate of core research field who obtains q gathers P ₁, screening patent quantity is no more than other patentees of setting threshold, and remaining patentee constitutes core research field candidate and gathers C ₁

The potential affiliate's recommend method of above-mentioned a kind of patentee, described step 2.2 comprises following substep:

Step 2.21 is collected the patent file of v and q and is gathered D _Qv={ d ₁, d ₂..., d _n;

Step 2.22, patent file is represented that with the space vector form concrete operation method is as follows:

For patent file set D _Qv={ d ₁, d ₂..., d _nIn any patent d _i, utilize the space vector of one group of keyword to represent; Its process is, at first all patent files carried out Chinese word segmentation, removes stop words in the document according to self-defined or public stop words dictionary then, for removing lexical item behind the stop words, calculates the weight of each lexical item in document, and its calculating formula is:

$w(t_j,d_i)=\frac{\mathrm{tf}(t_j,d_i)\×\log (N/n_{t_j}+0.01)}{\sqrt{{\mathrm{\Σ}}_{t_j\∈d_i}{[\mathrm{tf}(t_j,d_i)\×\log \left(N/n_{t_j}\right)+0.01]}^2}}$ Formula two;

Wherein, w (t _j, d _i) be lexical item t _jAt text d _iIn weight, and tf (t _j, d _i) be word t _jAt text d _iIn word frequency, N is patent set D _QvThe sum of middle patent, n are patent set D _QvThe patent file number of lexical item occurs, denominator is normalized factor;

At last, represent each piece patent file with the space vector of each lexical item correspondence, be expressed as

Wherein certain is worth w _IjBe lexical item t _jAt patent file d _iIn weight;

Step 2.23 is represented according to the space vector of document, calculates the similarity between the patent file in twos, and concrete operation method is:

To any two patent file d _iAnd d _j, use included angle cosine between the vector of its correspondence to come the similarity of measurement, its formula is:

$\mathrm{sim}(d_i,d_j)=\frac{\underset{l=1}{\overset{n}{\mathrm{\Σ}}}{w}_l\left(d_i\right)\×w_l\left(d_j\right)}{\sqrt{\left(\underset{l=1}{\overset{n}{\mathrm{\Σ}}}{w}_l^2\left(d_i\right)\right)\×\left(\underset{l=1}{\overset{n}{\mathrm{\Σ}}}{w}_l^2\left(d_j\right)\right)}}$ Formula three;

W wherein _l(d _i) be that l lexical item is at document d _iIn weight, w _l(d _j) be that l lexical item is at document d _jIn weight;

Step 2.24, according to the similarity that step 2.23 is calculated, to the patent text cluster, formed each class i.e. a technical theme, produces the theme S set _Qv

The potential affiliate's recommend method of above-mentioned a kind of patentee, described step 2.3 comprises following substep:

Step 2.31, according to the cluster result of step 2.24, calculate q and the complementary degree of v on any two technical theme i inequality and j:

$g(q,v,i,j)=\left\{\begin{array}{cc}\frac{n_{\mathrm{qi}}-n_{\mathrm{vi}}}{{\mathrm{sum}}_i}\×\frac{n_{\mathrm{vj}}-n_{\mathrm{qj}}}{{\mathrm{sum}}_j}& \mathrm{if}(n_{\mathrm{qi}}-n_{\mathrm{vi}})\×(n_{\mathrm{vj}}-n_{\mathrm{qj}})>0\\ 0& \mathrm{else}\end{array}\right.$ Formula four;

Wherein, n _QiBe the patent quantity of q in theme i, n _ViBe the patent quantity of v in theme i, sum _iBe the theme patent sum in the i, n _QjBe the patent quantity of q in theme j, n _VjBe the patent quantity of v in theme j, sum _jThe interior patent sum of j is the theme;

Step 2.32, according to the result of calculation of step 2.31, calculate the technology complementation degree between q and v:

$C(q,v)=\underset{i,j\∈S_{\mathrm{qv}}}{\mathrm{\Σ}}g(q,v,i,j)$ Formula five;

Wherein, S _QvBe the theme set that cluster forms, i, j are two different technical themes.

The potential affiliate's recommend method of above-mentioned a kind of patentee, described step 2.4 comprises following substep:

Step 2.41, v is in S set for statistics ₂Under patent quantity and application time, research and develop strength, calculating formula in the field according to patent quantity and application time calculated candidate patentee:

$P\left(v\right)=\underset{d\∈S}{\mathrm{\Σ}}{e}^{-\frac{t-t_d}{\mathrm{\λ}}}$ Formula six;

Wherein, S represents that v is in S set ₂Under patent file set, t _dBe the application time of patent file d, t is the current time, and λ is adjustable parameter;

Step 2.42, v is in S set for statistics ₂Under inventor's quantity, affiliate's quantity;

Step 2.43, calculate the possibility that v becomes the q affiliate, calculating formula:

Candidate (v)=and α C (q, v)+β P (v)+γ n ₁+ τ n ₂Formula seven;

Wherein, α, β, γ, τ are adjustable parameters, and v is the candidate patentee, n ₁Be that v is in S set ₂Inventor's quantity, n ₂Be that v is in S set ₂Affiliate's quantity.

The potential affiliate's recommend method of above-mentioned a kind of patentee, described step 2.5 concrete operation method is:

To the result of calculation of step 2.43, descending sort, K patentee is as the potential affiliate of the core realm of q before getting.

The potential affiliate's recommend method of above-mentioned a kind of patentee, the concrete operation method of described step 3.1 is:

Analyze S set ₃In patentee under each IPC, if patentee one hurdle of the patent file of q not only exists q to also have other patentees, then this patentee is considered as having cooperative relationship with q, the existing affiliate of non-core research field who obtains q gathers P ₂, other patentees constitute non-core research field candidate and gather C ₂

The potential affiliate's recommend method of above-mentioned a kind of patentee, the concrete operation method of described step 3.2 is:

Collect q and set P ₂In patentee k in S set ₃All enjoy Patent right patent file S set jointly down _Qk, calculate their cooperative relationship intensity again:

$R(q,k)=\underset{d\∈S_{\mathrm{qk}}}{\mathrm{\Σ}}{e}^{-\left(\frac{t-t_d}{t}\right)}$ Formula eight;

Wherein, t is the current time, t _dBe S _QkIn the application time of patent file d in the set, λ is adjustable parameter.

The potential affiliate's recommend method of above-mentioned a kind of patentee, described step 3.3 comprises following substep:

Step 3.31 is represented patentee's vectorization based on the thought of this paper vectorization, represents the patentee with the n-dimensional vector that keyword is formed, with c and k in S set ₃Under the set of all patent files be considered as the super document d of one piece of expression c and k _cAnd d _k, c and set C ₃In all patentees' super document constitute collection of document D _Sim={ d ₁, d ₂..., d _n, for patent file set D _Sim={ d ₁, d ₂..., d _nIn any patent d _iUtilize the space vector of one group of keyword to represent: its process is, at first patent file is carried out Chinese word segmentation, then according to the stop words in the self-defined or public stop words dictionary removal document, for removing lexical item behind the stop words, calculate the weight of each lexical item in document, its calculating formula is:

$w(t_j,d_i)=\frac{\mathrm{tf}(t_j,d_i)\×\log (N/n_{t_j}+0.01)}{\sqrt{{\mathrm{\Σ}}_{t_j\∈d_i}{[\mathrm{tf}(t_j,d_i)\×\log \left(N/n_{t_j}\right)+0.01]}^2}}$ Formula two;

Wherein, w (t _j, d _i) be lexical item t _jAt text d _iIn weight, and tf (t _j, d _i) be word t _jAt text d _iIn word frequency, N is patent set D _SimThe sum of middle patent, n are patent set D _SimThe patent file number of lexical item occurs, denominator is normalized factor;

At last, represent each piece patent file with the space vector of each lexical item correspondence, be expressed as Wherein certain is worth w _IjBe lexical item t _jAt patent file d _iIn weight;

Step 3.32 is to any two patent file d _iAnd d _j, use included angle cosine between the vector of its correspondence to come the similarity of measurement, its formula is:

$\mathrm{sim}(d_i,d_j)=\frac{\underset{l=1}{\overset{n}{\mathrm{\Σ}}}{w}_l\left(d_i\right)\×w_l\left(d_j\right)}{\sqrt{\left(\underset{l=1}{\overset{n}{\mathrm{\Σ}}}{w}_l^2\left(d_i\right)\right)\×\left(\underset{l=1}{\overset{n}{\mathrm{\Σ}}}{w}_l^2\left(d_j\right)\right)}}$ Formula three;

W wherein _l(d _i) be that l lexical item is at document d _iIn weight, w _l(d _j) be that l lexical item is at document d _jIn weight.

The potential affiliate's recommend method of above-mentioned a kind of patentee, the concrete operation method of described step 3.4 is:

According to technology similarity and the cooperative relationship prediction of strength time c of k and q and the cooperative relationship intensity of q of c and k, concrete calculating formula is as follows:

$P(q,c)={\mathrm{Max}}_{k\∈P_2}(R(q,k)\×\mathrm{sim}(k,c))$ Formula nine;

Wherein, q is given patentee, and k is the affiliate of q, and c is the candidate, P ₂Be existing affiliate's set of q, (q k) is q and k cooperative relationship intensity to R, and (k c) is the technology similarity of k and c to sim.

The potential affiliate's recommend method of above-mentioned a kind of patentee, described step 3.5 comprises following substep:

Step 3.51, c is in S set for statistics ₃Under patent quantity and application time, research and develop strength, calculating formula in the field according to patent quantity and application time calculated candidate patentee:

$P\left(c\right)=\underset{d\∈S}{\mathrm{\Σ}}{e}^{-\frac{t-t_d}{\mathrm{\λ}}}$ Formula six;

Step 3.52, c is in S set for statistics ₃Under inventor's quantity, affiliate's quantity;

Step 3.53, calculate the possibility that c becomes the q affiliate, calculating formula:

Candidate (c)=α C (q, v)+β P (v)+γ n ₁+ τ n ₂Formula seven;

Wherein, α, β, γ, τ are adjustable parameters, and c is the candidate patentee, n ₁Be that c is in S set ₃Under inventor's quantity, n ₂Be that c is in S set ₃Under affiliate's quantity.

The potential affiliate's recommend method of above-mentioned a kind of patentee, the concrete operation method of described step 3.6 is:

To the result of calculation of step 3.53, by descending sort, get preceding K patentee as the potential affiliate in non-core field of q.

Therefore, the present invention has following advantage: proposed the potential affiliate's discover method of patentee first 1.; 2. solved the technology complementation degree computational problem of patentee its core research field and other patentees; 3. solved vehicular with the existing affiliate cooperative relationship predictor calculation problem with other patentees in its non-core research field of patentee; 4. has the potential affiliate's recommendation effect of good patentee.

Description of drawings

The flow process that Fig. 1 recommends for the potential affiliate of patentee among the present invention.

Embodiment

Below by embodiment, and by reference to the accompanying drawings, technical scheme of the present invention is described in further detail.

Embodiment:

The potential affiliate's recommend method of a kind of patentee is characterized in that, based on definition: patent file set D={d ₁, d ₂..., d _n, the patentee gathers P={P ₁, P ₂..., P _m; Wherein, d _iThe document content of expression patent i; P _jRepresent j patentee; According to the relative patent quantity rank of given patentee under each IPC classification number that its patent relates to, the field of IPC classification number representative is classified as given patentee's core research field or non-core research field, the patentee of core research field recommended technology complementation, recommend the patentee similar to its existing affiliate's technology in non-core research field; Comprise and judge that research field is the step in core realm or non-core field, core research field given patentee is the step that given patentee recommends potential affiliate: and be the step that given patentee recommends potential affiliate in given patentee's non-core research field, specifically comprise:

Judge that research field is the step of core research field or non-core research field:

Step 1.1 is analyzed other IPC of group level under the patent of given patentee q, obtains the IPC S set ₁

Step 1.2, pair set S ₁In each IPC, the patent quantity of adding up all patentees under this IPC, by the descending sort of patent quantity to patentee's rank;

Step 1.3, calculate q at the relative patent quantity rank R (q) in this IPC representative field according to the ranking result of step 1.2, if R (q) surpasses setting threshold, then this field is the core research field of q, if R (q) does not surpass setting threshold, then this field is the non-core research field of q;

Step 1.4, pair set S ₁In after each IPC analyzes, obtain the core research field S set of q ₂With non-core research field S set ₃

Core research field given patentee is the step that given patentee recommends potential affiliate:

Step 2.1 is analyzed S set ₂In patentee under each IPC, if patentee one hurdle of the patent file of q not only exists q to also have other patentees, then this patentee is considered as having cooperative relationship with q, the existing affiliate of core research field who obtains q gathers P ₁, screening patent quantity is no more than other patentees of setting threshold, and remaining patentee constitutes core research field candidate and gathers C ₁

Step 2.2, pair set C ₁In each patentee v, collect it and q in S set ₂In patent file under each IPC, with the document vectorization, calculate the document similarity, according to similarity with clustering documents;

Step 2.3 according to the cluster result of step 2.2, is calculated the technology complementation degree of q and v;

Step 2.4 according to the result of calculation of step 2.3, is taken all factors into consideration v in S set ₂Under research and development strength, inventor's quantity, affiliate's quantity, calculate the possibility that v becomes the q affiliate;

Step 2.5, to the result of calculation of step 2.4, descending sort is got preceding K patentee as the potential affiliate of core research field of q;

Non-core research field given patentee is the step that given patentee recommends potential affiliate:

Step 3.1 is analyzed S set ₃In patentee under each IPC, if patentee one hurdle of the patent file of q not only exists q to also have other patentees, then this patentee is considered as having cooperative relationship with q, the existing affiliate of non-core research field who obtains q gathers P ₂, other patentees constitute non-core research field candidate and gather C ₂

Step 3.2, pair set P ₂In each patentee k, calculate the cooperative relationship intensity of k and q;

Step 3.3, pair set C ₂In each patentee c, calculate he with the set P ₂In each patentee's technology similarity;

Step 3.4 is according to the cooperative relationship intensity of step 3.2 calculating and the c and set P of step 3.3 calculating ₂In each patentee's technology similarity, calculate the cooperative relationship prediction of strength value of c and q;

Step 3.5 according to the result of calculation of step 3.4, is taken all factors into consideration c in S set ₃Under research and development strength, inventor's quantity, affiliate's quantity, calculate the possibility that c becomes the q affiliate;

Step 3.6, to the result of calculation of step 3.5, descending sort is got preceding K patentee as the potential affiliate in non-core field of q.

The potential affiliate's recommend method of above-mentioned a kind of patentee, described step 1.1 concrete operation method is:

Analyze affiliated other IPC of group level of patent of q, obtain the IPC S set ₁

The potential affiliate's recommend method of above-mentioned a kind of patentee, described step 1.2 concrete operation method is:

Pair set S ₁In each IPC, the patent quantity of adding up all patentees under this IPC, by the descending sort of patent quantity to patentee's rank.

The potential affiliate's recommend method of above-mentioned a kind of patentee, described step 1.3 concrete operation method is:

Calculate q at the relative patent quantity rank R (q) in this IPC representative field according to the ranking result of step 1.2, its calculating formula is:

$R\left(q\right)=\frac{\mathrm{Rank}\left(q\right)}{N}$ Formula one;

Wherein, Rank (q) is the rank of q, and N is all the patentee's quantity under this IPC;

If R (q) surpasses setting threshold, then this field is the core research field of q, if R (q) does not surpass setting threshold, then this field is the non-core research field of q.

The potential affiliate's recommend method of above-mentioned a kind of patentee, described step 1.4 concrete operation method is:

Pair set S ₁In after each IPC analyzes, obtain the core research field S set of q ₂With non-core research field S set ₃

The potential affiliate's recommend method of above-mentioned a kind of patentee, described step 2.1 concrete operation method is:

Analyze S set ₂In patentee under each IPC, if patentee one hurdle of the patent file of q not only exists q to also have other patentees, then this patentee is considered as having cooperative relationship with q, the existing affiliate of core research field who obtains q gathers P ₁, screening patent quantity is no more than other patentees of setting threshold, and remaining patentee constitutes core research field candidate and gathers C ₁

The potential affiliate's recommend method of above-mentioned a kind of patentee, described step 2.2 comprises following substep:

Step 2.21 is collected the patent file of v and q and is gathered D _Qv={ d ₁, d ₂..., d _n;

Step 2.22, patent file is represented that with the space vector form concrete operation method is as follows:

For patent file set D _Qv={ d ₁, d ₂..., d _nIn any patent d _i, utilize the space vector of one group of keyword to represent; Its process is, at first all patent files carried out Chinese word segmentation, removes stop words in the document according to self-defined or public stop words dictionary then, for removing lexical item behind the stop words, calculates the weight of each lexical item in document, and its calculating formula is:

$w(t_j,d_i)=\frac{\mathrm{tf}(t_j,d_i)\×\log (N/n_{t_j}+0.01)}{\sqrt{{\mathrm{\Σ}}_{t_j\∈d_i}{[\mathrm{tf}(t_j,d_i)\×\log \left(N/n_{t_j}\right)+0.01]}^2}}$ Formula two;

Wherein, w (t _j, d _i) be lexical item t _jAt text d _iIn weight, and tf (t _j, d _i) be word t _jAt text d _iIn word frequency, N is patent set D _QvThe sum of middle patent, n are patent set D _QvThe patent file number of lexical item occurs, denominator is normalized factor;

At last, represent each piece patent file with the space vector of each lexical item correspondence, be expressed as Wherein certain is worth w _IjBe lexical item t _jAt patent file d _iIn weight;

Step 2.23 is represented according to the space vector of document, calculates the similarity between the patent file in twos, and concrete operation method is:

To any two patent file d _iAnd d _j, use included angle cosine between the vector of its correspondence to come the similarity of measurement, its formula is:

$\mathrm{sim}(d_i,d_j)=\frac{\underset{l=1}{\overset{n}{\mathrm{\Σ}}}{w}_l\left(d_i\right)\×w_l\left(d_j\right)}{\sqrt{\left(\underset{l=1}{\overset{n}{\mathrm{\Σ}}}{w}_l^2\left(d_i\right)\right)\×\left(\underset{l=1}{\overset{n}{\mathrm{\Σ}}}{w}_l^2\left(d_j\right)\right)}}$ Formula three;

W wherein _l(d _i) be that l lexical item is at document d _iIn weight, w _l(d _j) be that l lexical item is at document d _jIn weight;

Step 2.24, according to the similarity that step 2.23 is calculated, to the patent text cluster, formed each class i.e. a technical theme, produces the theme S set _Qv

The potential affiliate's recommend method of above-mentioned a kind of patentee, described step 2.3 comprises following substep:

Step 2.31, according to the cluster result of step 2.24, calculate q and the complementary degree of v on any two technical theme i inequality and j:

$g(q,v,i,j)=\left\{\begin{array}{cc}\frac{n_{\mathrm{qi}}-n_{\mathrm{vi}}}{{\mathrm{sum}}_i}\×\frac{n_{\mathrm{vj}}-n_{\mathrm{qj}}}{{\mathrm{sum}}_j}& \mathrm{if}(n_{\mathrm{qi}}-n_{\mathrm{vi}})\×(n_{\mathrm{vj}}-n_{\mathrm{qj}})>0\\ 0& \mathrm{else}\end{array}\right.$ Formula four;

Wherein, n _QiBe the patent quantity of q in theme i, n _ViBe the patent quantity of v in theme i, sum _iBe the theme patent sum in the i, n _QjBe the patent quantity of q in theme j, n _VjBe the patent quantity of v in theme j, sum _jThe interior patent sum of j is the theme;

Step 2.32, according to the result of calculation of step 2.31, calculate the technology complementation degree between q and v:

$C(q,v)=\underset{i,j\∈S_{\mathrm{qv}}}{\mathrm{\Σ}}g(q,v,i,j)$ Formula five;

Wherein, S _QvBe the theme set that cluster forms, i, j are two different technical themes.

The potential affiliate's recommend method of above-mentioned a kind of patentee, described step 2.4 comprises following substep:

Step 2.41, v is in S set for statistics ₂Under patent quantity and application time, research and develop strength, calculating formula in the field according to patent quantity and application time calculated candidate patentee:

$P\left(v\right)=\underset{d\∈S}{\mathrm{\Σ}}{e}^{-\frac{t-t_d}{\mathrm{\λ}}}$ Formula six;

Wherein, S represents that v is in S set ₂Under patent file set, t _dBe the application time of patent file d, t is the current time, and λ is adjustable parameter;

Step 2.42, v is in S set for statistics ₂Under inventor's quantity, affiliate's quantity;

Step 2.43, calculate the possibility that v becomes the q affiliate, calculating formula:

Candidate (v)=and α C (q, v)+β P (v)+γ n ₁+ τ n ₂Formula seven;

Wherein, α, β, γ, τ are adjustable parameters, and v is the candidate patentee, n ₁Be that v is in S set ₂Inventor's quantity, n ₂Be that v is in S set ₂Affiliate's quantity.

The potential affiliate's recommend method of above-mentioned a kind of patentee, described step 2.5 concrete operation method is:

To the result of calculation of step 2.43, descending sort, K patentee is as the potential affiliate of the core realm of q before getting.

The potential affiliate's recommend method of above-mentioned a kind of patentee, the concrete operation method of described step 3.1 is:

Analyze S set ₃In patentee under each IPC, if patentee one hurdle of the patent file of q not only exists q to also have other patentees, then this patentee is considered as having cooperative relationship with q, the existing affiliate of non-core research field who obtains q gathers P ₂, other patentees constitute non-core research field candidate and gather C ₂

The potential affiliate's recommend method of above-mentioned a kind of patentee, the concrete operation method of described step 3.2 is:

Collect q and set P ₂In patentee k in S set ₃All enjoy Patent right patent file S set jointly down _Qk, calculate their cooperative relationship intensity again:

$R(q,k)=\underset{d\∈S_{\mathrm{qk}}}{\mathrm{\Σ}}{e}^{-\left(\frac{t-t_d}{t}\right)}$ Formula eight;

Wherein, t is the current time, t _dBe S _QkIn the application time of patent file d in the set, λ is adjustable parameter.

The potential affiliate's recommend method of above-mentioned a kind of patentee, described step 3.3 comprises following substep:

Step 3.31 is represented patentee's vectorization based on the thought of this paper vectorization, represents the patentee with the n-dimensional vector that keyword is formed, with c and k in S set ₃Under the set of all patent files be considered as the super document d of one piece of expression c and k _cAnd d _k, c and set C ₃In all patentees' super document constitute collection of document D _Sim={ d ₁, d ₂..., d _n, for patent file set D _Sim={ d ₁, d ₂..., d _nIn any patent d _iUtilize the space vector of one group of keyword to represent: its process is, at first patent file is carried out Chinese word segmentation, then according to the stop words in the self-defined or public stop words dictionary removal document, for removing lexical item behind the stop words, calculate the weight of each lexical item in document, its calculating formula is:

$w(t_j,d_i)=\frac{\mathrm{tf}(t_j,d_i)\×\log (N/n_{t_j}+0.01)}{\sqrt{{\mathrm{\Σ}}_{t_j\∈d_i}{[\mathrm{tf}(t_j,d_i)\×\log \left(N/n_{t_j}\right)+0.01]}^2}}$ Formula two;

Wherein, w (t _j, d _i) be lexical item t _jAt text d _iIn weight, and tf (t _j, d _i) be word t _jAt text d _iIn word frequency, N is patent set D _SimThe sum of middle patent, n are patent set D _SimThe patent file number of lexical item occurs, denominator is normalized factor;

At last, represent each piece patent file with the space vector of each lexical item correspondence, be expressed as Wherein certain is worth w _IjBe lexical item t _jAt patent file d _iIn weight;

Step 3.32 is to any two patent file d _iAnd d _j, use included angle cosine between the vector of its correspondence to come the similarity of measurement, its formula is:

$\mathrm{sim}(d_i,d_j)=\frac{\underset{l=1}{\overset{n}{\mathrm{\Σ}}}{w}_l\left(d_i\right)\×w_l\left(d_j\right)}{\sqrt{\left(\underset{l=1}{\overset{n}{\mathrm{\Σ}}}{w}_l^2\left(d_i\right)\right)\×\left(\underset{l=1}{\overset{n}{\mathrm{\Σ}}}{w}_l^2\left(d_j\right)\right)}}$ Formula three;

W wherein _l(d _i) be that l lexical item is at document d _iIn weight, w _l(d _j) be that l lexical item is at document d _jIn weight.

The potential affiliate's recommend method of above-mentioned a kind of patentee, the concrete operation method of described step 3.4 is:

According to technology similarity and the cooperative relationship prediction of strength time c of k and q and the cooperative relationship intensity of q of c and k, concrete calculating formula is as follows:

$P(q,c)={\mathrm{Max}}_{k\∈P_2}(R(q,k)\×\mathrm{sim}(k,c))$ Formula nine;

Wherein, q is given patentee, and k is the affiliate of q, and c is the candidate, P ₂Be existing affiliate's set of q, (q k) is q and k cooperative relationship intensity to R, and (k c) is the technology similarity of k and c to sim.

The potential affiliate's recommend method of above-mentioned a kind of patentee, described step 3.5 comprises following substep:

Step 3.51, c is in S set for statistics ₃Under patent quantity and application time, research and develop strength, calculating formula in the field according to patent quantity and application time calculated candidate patentee:

$P\left(c\right)=\underset{d\∈S}{\mathrm{\Σ}}{e}^{-\frac{t-t_d}{\mathrm{\λ}}}$ Formula six;

Step 3.52, c is in S set for statistics ₃Under inventor's quantity, affiliate's quantity;

Step 3.53, calculate the possibility that c becomes the q affiliate, calculating formula:

Candidate (c)=α C (q, v)+β P (v)+γ n ₁+ τ n ₂Formula seven;

Wherein, α, β, γ, τ are adjustable parameters, and c is the candidate patentee, n ₁Be that c is in S set ₃Under inventor's quantity, n ₂Be that c is in S set ₃Under affiliate's quantity.

The potential affiliate's recommend method of above-mentioned a kind of patentee, the concrete operation method of described step 3.6 is:

To the result of calculation of step 3.53, by descending sort, get preceding K patentee as the potential affiliate in non-core field of q.

What should specify is: the present invention has following main beneficial effect: the patentee who proposes among the present invention potential affiliate's recommend method has remedied the blank that the potential affiliate of patentee recommends research field; The potential affiliate's recommend method of the patentee who proposes among the 2nd, the present invention has good validity, specifies as follows:

The present invention is based on the patentee can be in the hypothesis of prior art field continuation research, and at first the research field with the patentee is classified as core research field or non-core research neck.Considered that the patentee seeks the patent cooperation partner that can offer help to its technical field in its core research field, has considered that also the patentee seeks the patent cooperation partner to having affiliate's technology now similar in its non-core research field.Estimate the possibility that given patentee affiliate potential with it cooperates with regard to two kinds of situations respectively.

Specific embodiment described herein only is that the present invention's spirit is illustrated.Those skilled in the art can make various modifications or replenish or adopt similar mode to substitute described specific embodiment, but can't depart from spirit of the present invention or surmount the defined scope of appended claims.

Claims (16)

1. the potential affiliate's recommend method of patentee is characterized in that, based on definition: patent file set D={d ₁, d ₂..., d _n, the patentee gathers P={P ₁, P ₂..., P _m; Wherein, d _iThe document content of expression patent i; P _jRepresent j patentee; According to the relative patent quantity rank of given patentee under each IPC classification number that its patent relates to, the field of IPC classification number representative is classified as given patentee's core research field or non-core research field, the patentee of core research field recommended technology complementation, recommend the patentee similar to its existing affiliate's technology in non-core research field; Comprise and judge that research field is the step in core realm or non-core field, core research field given patentee is the step that given patentee recommends potential affiliate: and be the step that given patentee recommends potential affiliate in given patentee's non-core research field, specifically comprise:

Judge that research field is the step of core research field or non-core research field:

Step 1.1 is analyzed other IPC of group level under the patent of given patentee q, obtains the IPC S set ₁

Step 1.2, pair set S ₁In each IPC, the patent quantity of adding up all patentees under this IPC, by the descending sort of patent quantity to patentee's rank;

Step 1.3, calculate q at the relative patent quantity rank R (q) in this IPC representative field according to the ranking result of step 1.2, if R (q) surpasses setting threshold, then this field is the core research field of q, if R (q) does not surpass setting threshold, then this field is the non-core research field of q;

Step 1.4, pair set S ₁In after each IPC analyzes, obtain the core research field S set of q ₂With non-core research field S set ₃

Core research field given patentee is the step that given patentee recommends potential affiliate:

Step 2.1 is analyzed S set ₂In patentee under each IPC, if patentee one hurdle of the patent file of q not only exists q to also have other patentees, then this patentee is considered as having cooperative relationship with q, the existing affiliate of core research field who obtains q gathers P ₁, screening patent quantity is no more than other patentees of setting threshold, and remaining patentee constitutes core research field candidate and gathers C ₁

Step 2.2, pair set C ₁In each patentee v, collect it and q in S set ₂In patent file under each IPC, with the document vectorization, calculate the document similarity, according to similarity with clustering documents;

Step 2.3 according to the cluster result of step 2.2, is calculated the technology complementation degree of q and v;

Step 2.4 according to the result of calculation of step 2.3, is taken all factors into consideration v in S set ₂Under research and development strength, inventor's quantity, affiliate's quantity, calculate the possibility that v becomes the q affiliate;

Step 2.5, to the result of calculation of step 2.4, descending sort is got preceding K patentee as the potential affiliate of core research field of q;

Non-core research field given patentee is the step that given patentee recommends potential affiliate:

Step 3.1 is analyzed S set ₃In patentee under each IPC, if patentee one hurdle of the patent file of q not only exists q to also have other patentees, then this patentee is considered as having cooperative relationship with q, the existing affiliate of non-core research field who obtains q gathers P ₂, other patentees constitute non-core research field candidate and gather C ₂

Step 3.2, pair set P ₂In each patentee k, calculate the cooperative relationship intensity of k and q;

Step 3.3, pair set C ₂In each patentee c, calculate he with the set P ₂In each patentee's technology similarity;

Step 3.4 is according to the cooperative relationship intensity of step 3.2 calculating and the c and set P of step 3.3 calculating ₂In each patentee's technology similarity, calculate the cooperative relationship prediction of strength value of c and q;

Step 3.5 according to the result of calculation of step 3.4, is taken all factors into consideration c in S set ₃Under research and development strength, inventor's quantity, affiliate's quantity, calculate the possibility that c becomes the q affiliate;

Step 3.6, to the result of calculation of step 3.5, descending sort is got preceding K patentee as the potential affiliate in non-core field of q.

2. the potential affiliate's recommend method of a kind of patentee according to claim 1 is characterized in that, described step 1.1 concrete operation method is:

Analyze affiliated other IPC of group level of patent of q, obtain the IPC S set ₁

3. the potential affiliate's recommend method of a kind of patentee according to claim 1 is characterized in that, described step 1.2 concrete operation method is:

Pair set S ₁In each IPC, the patent quantity of adding up all patentees under this IPC, by the descending sort of patent quantity to patentee's rank.

4. the potential affiliate's recommend method of a kind of patentee according to claim 1 is characterized in that, described step 1.3 concrete operation method is:

Calculate q at the relative patent quantity rank R (q) in this IPC representative field according to the ranking result of step 1.2, its calculating formula is:

$R\left(q\right)=\frac{\mathrm{Rank}\left(q\right)}{N}$ Formula one;

Wherein, Rank (q) is the rank of q, and N is all the patentee's quantity under this IPC;

If R (q) surpasses setting threshold, then this field is the core research field of q, if R (q) does not surpass setting threshold, then this field is the non-core research field of q.

5. the potential affiliate's recommend method of a kind of patentee according to claim 1 is characterized in that, described step 1.4 concrete operation method is:

Pair set S ₁In after each IPC analyzes, obtain the core research field S set of q ₂With non-core research field S set ₃

6. the potential affiliate's recommend method of a kind of patentee according to claim 1 is characterized in that, described step 2.1 concrete operation method is:

Analyze S set ₂In patentee under each IPC, if patentee one hurdle of the patent file of q not only exists q to also have other patentees, then this patentee is considered as having cooperative relationship with q, the existing affiliate of core research field who obtains q gathers P ₁, screening patent quantity is no more than other patentees of setting threshold, and remaining patentee constitutes core research field candidate and gathers C ₁

7. the potential affiliate's recommend method of a kind of patentee according to claim 1 is characterized in that described step 2.2 comprises following substep:

Step 2.21 is collected the patent file of v and q and is gathered D _Qv={ d ₁, d ₂..., d _n;

Step 2.22, patent file is represented that with the space vector form concrete operation method is as follows:

For patent file set D _Qv={ d ₁, d ₂..., d _nIn any patent d _i, utilize the space vector of one group of keyword to represent; Its process is, at first all patent files carried out Chinese word segmentation, removes stop words in the document according to self-defined or public stop words dictionary then, for removing lexical item behind the stop words, calculates the weight of each lexical item in document, and its calculating formula is:

$w(t_j,d_i)=\frac{\mathrm{tf}(t_j,d_i)\×\log (N/n_{t_j}+0.01)}{\sqrt{{\mathrm{\Σ}}_{t_j\∈d_i}{[\mathrm{tf}(t_j,d_i)\×\log \left(N/n_{t_j}\right)+0.01]}^2}}$ Formula two;

Wherein, w (t _j, d _i) be lexical item t _jAt text d _iIn weight, and tf (t _j, d _i) be word t _jAt text d _iIn word frequency, N is patent set D _QvThe sum of middle patent, n are patent set D _QvThe patent file number of lexical item occurs, denominator is normalized factor;

At last, represent each piece patent file with the space vector of each lexical item correspondence, be expressed as

Wherein certain is worth w _IjBe lexical item t _jAt patent file d _iIn weight;

Step 2.23 is represented according to the space vector of document, calculates the similarity between the patent file in twos, and concrete operation method is:

To any two patent file d _iAnd d _j, use included angle cosine between the vector of its correspondence to come the similarity of measurement, its formula is:

$\mathrm{sim}(d_i,d_j)=\frac{\underset{l=1}{\overset{n}{\mathrm{\Σ}}}{w}_l\left(d_i\right)\×w_l\left(d_j\right)}{\sqrt{\left(\underset{l=1}{\overset{n}{\mathrm{\Σ}}}{w}_l^2\left(d_i\right)\right)\×\left(\underset{l=1}{\overset{n}{\mathrm{\Σ}}}{w}_l^2\left(d_j\right)\right)}}$ Formula three;

W wherein _l(d _i) be that l lexical item is at document d _iIn weight, w _l(d _j) be that l lexical item is at document d _jIn weight;

Step 2.24, according to the similarity that step 2.23 is calculated, to the patent text cluster, formed each class i.e. a technical theme, produces the theme S set _Qv

8. the potential affiliate's recommend method of a kind of patentee according to claim 1 is characterized in that described step 2.3 comprises following substep:

Step 2.31, according to the cluster result of step 2.24, calculate q and the complementary degree of v on any two technical theme i inequality and j:

$g(q,v,i,j)=\left\{\begin{array}{cc}\frac{n_{\mathrm{qi}}-n_{\mathrm{vi}}}{{\mathrm{sum}}_i}\×\frac{n_{\mathrm{vj}}-n_{\mathrm{qj}}}{{\mathrm{sum}}_j}& \mathrm{if}(n_{\mathrm{qi}}-n_{\mathrm{vi}})\×(n_{\mathrm{vj}}-n_{\mathrm{qj}})>0\\ 0& \mathrm{else}\end{array}\right.$ Formula four;

Wherein, n _QiBe the patent quantity of q in theme i, n _ViBe the patent quantity of v in theme i, sum _iBe the theme patent sum in the i, n _QjBe the patent quantity of q in theme j, n _VjBe the patent quantity of v in theme j, sum _jThe interior patent sum of j is the theme;

Step 2.32, according to the result of calculation of step 2.31, calculate the technology complementation degree between q and v:

$C(q,v)=\underset{i,j\∈S_{\mathrm{qv}}}{\mathrm{\Σ}}g(q,v,i,j)$ Formula five;

Wherein, S _QvBe the theme set that cluster forms, i, j are two different technical themes.

9. the potential affiliate's recommend method of a kind of patentee according to claim 1 is characterized in that described step 2.4 comprises following substep:

Step 2.41, v is in S set for statistics ₂Under patent quantity and application time, research and develop strength, calculating formula in the field according to patent quantity and application time calculated candidate patentee:

$P\left(v\right)=\underset{d\∈S}{\mathrm{\Σ}}{e}^{-\frac{t-t_d}{\mathrm{\λ}}}$ Formula six;

Wherein, S represents that v is in S set ₂Under patent file set, t _dBe the application time of patent file d, t is the current time, and λ is adjustable parameter;

Step 2.42, v is in S set for statistics ₂Under inventor's quantity, affiliate's quantity;

Step 2.43, calculate the possibility that v becomes the q affiliate, calculating formula:

Candidate (v)=and α C (q, v)+β P (v)+γ n ₁+ τ n ₂Formula seven;

Wherein, α, β, γ, τ are adjustable parameters, and v is the candidate patentee, n ₁Be that v is in S set ₂Inventor's quantity, n ₂Be that v is in S set ₂Affiliate's quantity.

10. the potential affiliate's recommend method of a kind of patentee according to claim 9 is characterized in that, described step 2.5 concrete operation method is:

To the result of calculation of step 2.43, descending sort, K patentee is as the potential affiliate of the core realm of q before getting.

11. the potential affiliate's recommend method of a kind of patentee according to claim 1 is characterized in that the concrete operation method of described step 3.1 is:

Analyze S set ₃In patentee under each IPC, if patentee one hurdle of the patent file of q not only exists q to also have other patentees, then this patentee is considered as having cooperative relationship with q, the existing affiliate of non-core research field who obtains q gathers P ₂, other patentees constitute non-core research field candidate and gather C ₂

12. the potential affiliate's recommend method of a kind of patentee according to claim 1 is characterized in that the concrete operation method of described step 3.2 is:

Collect q and set P ₂In patentee k in S set ₃All enjoy Patent right patent file S set jointly down _Qk, calculate their cooperative relationship intensity again:

$R(q,k)=\underset{d\∈S_{\mathrm{qk}}}{\mathrm{\Σ}}{e}^{-\left(\frac{t-t_d}{t}\right)}$ Formula eight;

Wherein, t is the current time, t _dBe S _QkIn the application time of patent file d in the set, λ is adjustable parameter.

13. the potential affiliate's recommend method of a kind of patentee according to claim 1 is characterized in that described step 3.3 comprises following substep:

Step 3.31 is represented patentee's vectorization based on the thought of this paper vectorization, represents the patentee with the n-dimensional vector that keyword is formed, with c and k in S set ₃Under the set of all patent files be considered as the super document d of one piece of expression c and k _cAnd d _k, c and set C ₃In all patentees' super document constitute collection of document D _Sim={ d ₁, d ₂..., d _n, for patent file set D _Sim={ d ₁, d ₂..., d _nIn any patent d _iUtilize the space vector of one group of keyword to represent: its process is, at first patent file is carried out Chinese word segmentation, then according to the stop words in the self-defined or public stop words dictionary removal document, for removing lexical item behind the stop words, calculate the weight of each lexical item in document, its calculating formula is:

$w(t_j,d_i)=\frac{\mathrm{tf}(t_j,d_i)\×\log (N/n_{t_j}+0.01)}{\sqrt{{\mathrm{\Σ}}_{t_j\∈d_i}{[\mathrm{tf}(t_j,d_i)\×\log \left(N/n_{t_j}\right)+0.01]}^2}}$ Formula two;

Wherein, w (t _j, d _i) be lexical item t _jAt text d _iIn weight, and tf (t _j, d _i) be word t _jAt text d _iIn word frequency, N is patent set D _SimThe sum of middle patent, n are patent set D _SimThe patent file number of lexical item occurs, denominator is normalized factor;

At last, represent each piece patent file with the space vector of each lexical item correspondence, be expressed as

Wherein certain is worth w _IjBe lexical item t _jAt patent file d _iIn weight;

Step 3.32 is to any two patent file d _iAnd d _j, use included angle cosine between the vector of its correspondence to come the similarity of measurement, its formula is:

$\mathrm{sim}(d_i,d_j)=\frac{\underset{l=1}{\overset{n}{\mathrm{\Σ}}}{w}_l\left(d_i\right)\×w_l\left(d_j\right)}{\sqrt{\left(\underset{l=1}{\overset{n}{\mathrm{\Σ}}}{w}_l^2\left(d_i\right)\right)\×\left(\underset{l=1}{\overset{n}{\mathrm{\Σ}}}{w}_l^2\left(d_j\right)\right)}}$ Formula three;

W wherein _l(d _i) be that l lexical item is at document d _iIn weight, w _l(d _j) be that l lexical item is at document d _jIn weight.

14. the potential affiliate's recommend method of a kind of patentee according to claim 1 is characterized in that the concrete operation method of described step 3.4 is:

According to technology similarity and the cooperative relationship prediction of strength time c of k and q and the cooperative relationship intensity of q of c and k, concrete calculating formula is as follows:

$P(q,c)={\mathrm{Max}}_{k\∈P_2}(R(q,k)\×\mathrm{sim}(k,c))$ Formula nine;

Wherein, q is given patentee, and k is the affiliate of q, and c is the candidate, P ₂Be existing affiliate's set of q, (q k) is q and k cooperative relationship intensity to R, and (k c) is the technology similarity of k and c to sim.

15. the potential affiliate's recommend method of a kind of patentee according to claim 1 is characterized in that described step 3.5 comprises following substep:

Step 3.51, c is in S set for statistics ₃Under patent quantity and application time, research and develop strength, calculating formula in the field according to patent quantity and application time calculated candidate patentee:

$P\left(c\right)=\underset{d\∈S}{\mathrm{\Σ}}{e}^{-\frac{t-t_d}{\mathrm{\λ}}}$ Formula six;

Step 3.52, c is in S set for statistics ₃Under inventor's quantity, affiliate's quantity;

Step 3.53, calculate the possibility that c becomes the q affiliate, calculating formula:

Candidate (c)=α C (q, v)+β P (v)+γ n ₁+ τ n ₂Formula seven;

Wherein, α, β, γ, τ are adjustable parameters, and c is the candidate patentee, n ₁Be that c is in S set ₃Under inventor's quantity, n ₂Be that c is in S set ₃Under affiliate's quantity.

16. the potential affiliate's recommend method of a kind of patentee according to claim 15 is characterized in that the concrete operation method of described step 3.6 is:

To the result of calculation of step 3.53, by descending sort, get preceding K patentee as the potential affiliate in non-core field of q.

Images