CN104112170A - Constructing method of patent leading indicator and application - Google Patents

Abstract

The invention provides a constructing method of a patent leading indicator and application. The method comprises: (1), setting a plurality of patent entities and a plurality of patent indicators for describing all patent entities and financial indicators; (2), setting a data collection period; (3), collecting patent indicator data and financial indicator data of the patent entities during the data collection period; (4), combining all patent indicator data and financial indicator data to form first panel data; (5), carrying out conversion operation program processing on the first panel data to form second panel data; (6), setting a time leading period and providing a time leading period-based first time sequence operation program, whose independent variables serve as patent indicator data of the second panel data and dependent variables serve as financial indicator data of the second panel data; and (7), setting a first threshold value and operating the second panel data according to the first time sequence operation program, and carrying out screening on the plurality of patent indicators to obtain a patent leading indicator meeting the first threshold value.

Description

The constructing method of patent leading indicators and application

[technical field]

The present invention relates to digital content treatment technology, refer in particular to by the computing of patent information and excavate the association of patent information to corporate financial information, and then set up Forecasting Methodology and the relative computer system of the leading financial information of patent information.

[background technology]

Along with the fast development of scientific and technological industry and the extremely attention of intellecture property, patent has been regarded as the important indicator of industry or technology.For obligee, patent has been not only the defence instrument for protection and creation and product, more becomes the optimal attack weapon of competing on industry stage with rival.If company can have more patents, just more representative and impact property on its competitive field, just therefore patent becomes a very important competitive power information.

World Intellectual Property Organization (WIPO) been reported, the technology contents that patent comprises, has 80% in other papers, magazine, encyclopedia, not disclose.Patent is the concrete manifestation of innovation.Patent numbers and quality are compared with the unit of tool advantage, and the energy of its innovation and quality also have more advantage compared with other rivals.Because patent possesses jural exclusive right, market is there is to the effect of oligopoly.Therefore for taking technical research for basic scientific and technological enterprises, patent numbers and quality are compared with tool dominant, its production marketing and achievement also possess the superiority of certain degree.

In prior art, many paper studies are pointed out, the sales information of the leading product of quantity information of patent, for example, for explaining that market development situation has the effect of primacy, following document:

(1)Griliches,Z.(1720),Patent?statistics?as?economic?indicators：a?survey,Journal?of?Economic?Literature,28(4),PP.1661-1707.

(2)Ernst,H.(1725),Patenting?strategies?in?the?German?mechanical?engineering?industry?and?their?relationship?to?firm?Performance,Technovation,5(4),PP.225-240.

(3)Adams,K.,D.Kim,F.L.JoLz,Trost,R.P.and?Mastrogianis,G.(1727),Modeling?and?Forecasting?U.S.Patent?Application?Filings,Journal?of?Policy?Modeling,19(5),PP.491-535.

(4)Ernst?H.(1727),The?Use?of?Patent?Data?for?Technological?Forecasting：The?Diffusion?of?CNC-Technology?in?Machine?Tool?Industry,Small?Business?Economics,9(4),PP.361-381.

(5)Ernst?H.(2001),Patent?applications?and?subsequent?changes?of?Performance：evidence?from?time-series?cross-section?analyses?on?the?firm?level,Research?Policy,30(1),PP.143-157.

Above-mentioned prior art points out that patent information becomes more important, more merits attention compared to other market informations, therefore with the method for patent information prediction markets information, just becomes gradually an important research topic in investment and valuation of enterprise.

In prior art, US Patent No. 6175824 proposes with patent information, Company Financial to be showed and assessed with US6832171 first.US Patent No. 6175824 and US6832171 are for the more high-tech stock of listed company's patent number, by multiple regression analysis model, analyze the association of listed company's patent index over the years (PI) and financial index (FI) over the years, finally derive taking patent index (PI) as basic Estimate equation formula, calculate its Value of Science & Technology with this Estimate equation formula, again with relatively its HSBC (Market-to-Book Ratio) of this innovative value, if being greater than HSBC, Value of Science & Technology is considered as investment potential, if being less than HSBC, Value of Science & Technology is considered as without investment value, the instrument of selecting stocks as investment by this.Its patentee CHI Research more by this Estimate equation formula based on patent information push masses to, set up accordingly the commercial pattern of its exclusive business.

Prior art US6175824 and US6832171 have its uniqueness, but shortcoming is that it analyzes patent index (PI) is typical multiple regression analysis model with the equation of financial index (FI), independent variable is a seasonal effect in time series patent index (PI), and dependent variable is the financial index (FI) of same time sequence.Therefore set up multiple regression equation, when input certain time point patent index (PI) time, the Output rusults obtaining is the corresponding financial index of this time point (FI), instead of " future " and financial index (FI).

In prior art, US Patent No. 6556992, US7657476 and US7716226, taking the sustainment rate of patent as basis, other patent indexes (PI) of arranging in pairs or groups again, develop the another set of appraisal procedure to listed company's innovation ability with multiple regression analysis model, and selected according to this 300 listed companies of the U.S. as constituent stocks, after weighted calculation, issue first patent index of the whole world (OT300Patent Index).But the multiple regression analysis model that US6556992, US7657476 and US7716226 use, its independent variable patent index (PI) is all to belong to same time sequence with dependent variable financial index (FI), although its independent variable is different from prior art US6175824 and US6832171, but the multiple regression equation of its foundation, in the time inputting the patent index (PI) of certain time point, the Output rusults obtaining remains the corresponding financial index of this time point (FI), instead of " future " and financial index (FI).

We it must be understood that, the Practical Operation that investment is selected stocks, and investment institution, in the time of investment, and non-hopely just makes a profit instantly, but wishes just to make a profit now at certain time point cover in future.That is to say, investment institution is in when investment, wish the information instantly grasped to " future " and profit have " prediction " and effect, could reduce investment risk, guarantee returns of investment.Above-mentioned prior art US6175824, US6832171, US6556992, US7657476 and US7716226 substantially do not have " prediction " effect.

Another aspect, above-mentioned prior art US6175824, US6832171, US6556992, US7657476 and US7716226 are information and the Index Establishment models having based on U.S.'s invention granted patent information.The most crucial patent index (PI) that wherein US6175824 and US6832171 use is examined the patent quoted passage that uses and the quoted passage of non-patent literature for novelty (NoveLy) examination and creativeness/unobviousness (Nonobviousness), comprises front quoted passage (Backward Citation) and rear quoted passage (Forward Citation); The most crucial patent index (PI) that US6556992, US7657476 and US7716226 use is the sustainment rate of invention granted patent.But United States Patent (USP) does not have utility model system, but vast patent information for China's Mainland, utility model patent that need not entity examination procedure is more, and patent of invention is less; For patent of invention, the granted patent of examining by entity is less, and early stage publication is more.For invention granted patent, the citation information that patent database is issued is that the granted patent of recent years just starts to disclose gradually, does not fully date back to previous granted patent again; For citation information, only have at present front quoted passage again, patent entity (PE) when examination the documents that adopts, but and unexposed rear quoted passage, after license, set it as documents and other patents of examining.So the content and method of prior art US6175824, US6832171, US6556992, US7657476 and its exposure of US7716226, the patent information of inapplicable China's Mainland, therefore cannot assess the financial information of the listed company of China's Mainland.

Edge this, for the financial data of the listed company of China's Mainland and in the patent information of China's Mainland, possess if having " prediction " effect association model and realize the computer system of this association model, not only contribute to the technical strength development of patent information analysis and utilization, the front development of capitalized method that more can investment promotion field, and the research and development of industrial technology and innovation are played to positive support effect.

[summary of the invention]

Based on improving above-mentioned defect of the prior art, fundamental purpose of the present invention is to provide a kind of constructing method (100) of patent leading indicators, patent leading indicators (172) are in order to predict the financial information of patent entity (PE), and the information of the financial index (FI) of the leading patent entity of the information of patent leading indicators (172) (PE) has the leading phase of predefined time (L), the constructing method (100) of patent leading indicators comprises:

(1) set multiple patent entities (PE) and multiple patent indexes (PI) and financial index (FI) in order to describe each patent entity (PE), each patent index (PI) is obtained by the patent information computing of each patent entity (PE).

(2) setting data is collected the phase (121), and collection period (121) is made up of time interval (T) and time issue (N), and time issue (N) is for being not less than two integer;

(3) collect each patent entity (PE) in collection period (121), the corresponding patent index data (131) of each time interval (T) institute and financial index data (132);

(4) by multiple patent index data (131) of multiple patent entities (PE) and multiple financial index data (132) composition the first panel datas (141);

(5) the first panel data (141) is formed to second panel data (151) of normal distribution and criterion score by translation operation program (152);

(6) leading phase of setting-up time (L) and the very first time Sequence Operation Theory program (161) based on the leading phase of time (L) is provided, the leading phase of time (L) comprises at least one time interval (T), the independent variable of very first time Sequence Operation Theory program (161) is patent index data (131) of the second panel data (151), and the dependent variable of very first time Sequence Operation Theory program (161) is the financial index data (132) of the second panel data (151);

(7) set first threshold (171), by very first time Sequence Operation Theory program (161) and leading phase of time (L), computing the second panel data (151), from multiple patent indexes (PI), screening draws at least one the patent leading indicators (172) that meets first threshold (171).

The constructing method (100) of patent leading indicators proposed by the invention is objective and rigorous, is not only particularly suitable for China's Mainland patent information, comprises disclosure of the invention patent, invention granted patent, utility model patent and design patent; Also be applicable to for other various countries' patent information simultaneously, excavate the patent leading indicators (172) with leading corporate financial information.

Another object of the present invention is to provide the leading equational constructing method of a kind of patent (500), patent leading side formula (501) is in order to predict the financial information of patent entity (PE), patent leading side formula (501) generates a leading mark of patent (502), and the financial information of the leading patent entity of the leading mark of patent (502) (PE) has the leading phase of predefined time (L).The constructing method (500) of patent leading side formula (501) comprising:

(1) obtain the second panel data (151) and multiple patent leading indicators (172), the second panel data (151) is obtained by the constructing method (100) of aforesaid patent leading indicators with multiple patent leading indicators (172).

(2), based on multiple patent leading indicators (172), form the 3rd panel data (521) from the second panel data (151) screening;

(3) provide the second time series operation program (531) based on the leading phase of time (L), the independent variable of the second time series operation program (531) is multiple patent leading indicators (172) of the 3rd panel data (521), and the dependent variable of the second time series operation program (531) is the financial index (FI) of the second panel data (151);

(4) set Second Threshold (541), by the second time series operation program (531) and leading phase of time (L), computing the 3rd panel data (521), from multiple patent leading indicators (172), screening draws and meets multiple patent core index (542) of Second Threshold (541) and generate patent leading side formula (501), and patent leading side formula (501) is made up of multiple patent core index (542) and corresponding weight coefficient (543) thereof in fact.

Another object of the present invention is to provide a kind of method (600) of assessing enterprise investment potentiality, comprising:

(1) collect the patent information (612) of multiple enterprises (611);

(2) provide a patent leading side formula (501), patent leading side formula (501) is obtained by the leading equational constructing method of aforementioned patent (500), and patent leading side formula (501) is made up of multiple patent core index (542) and corresponding weight coefficient (543) thereof in fact;

(3) patent information based on each enterprise (611) (612) is calculated the data (631) of the corresponding patent core index of each enterprise (542);

(4) data (631) of the patent core index (542) based on each enterprise (611), calculate the leading mark of patent (502) that generates each enterprise (611) by leading side formula (501);

(5) by a sequencer program (651), the leading mark of patent (502) of each enterprise (611) is sorted, ranking results (652) represents the investment potential of enterprise (611).

The higher person of the leading mark of above-mentioned patent (502), represent enterprise after the leading phase of time (L) institute the data of financial index (FI) are higher accordingly; The lower person of the leading mark of patent (502), represent enterprise after the leading phase of time (L) institute the numerical value of financial index (FI) is lower accordingly.Because the data height of business finance index (FI) is directly expressed the quality of its business performance, the higher person of numerical value of financial index (FI), performance better, more has investment value.Because the leading mark of patent (502) represents enterprise's institute's numerical value of financial index (FI) accordingly after the leading phase of time (L), therefore the height rank of the leading mark of the patent being had by enterprise (502), just can be from wherein picking out the object with investment potential.

A further object of the present invention is to provide a kind of computer system (700) of assessing enterprise investment potentiality, comprises patent information harvester (710), index calculation element (720), the leading mark calculation element of patent (730) and mark collator (740).Wherein, patent information harvester (710) is collected the patent information (612) of multiple enterprises; The patent information (612) of index calculation element (720) based on each enterprise calculated, and generates the data (631) of the corresponding patent core index of enterprise (542); The leading mark calculation element of patent (730) is according to the data (631) of the patent core index (542) of each enterprise, calculate the leading mark of patent (502) that generates each enterprise by patent leading side formula (501), patent leading side formula (501) is obtained by the leading equational constructing method of aforesaid patent (500), and patent leading side formula (501) is made up of multiple patent core index (542) and corresponding weight coefficient (543) thereof in fact; Mark collator (740) sorts leading multiple patents mark (502), and ranking results (652) represents the investment potential of multiple enterprises.

The method (600) of assessment enterprise investment potentiality proposed by the invention and computer system (700), it is the achievement based on large data, objective computing, rigorous checking, not only contribute to the technical strength development of patent information analysis and utilization, the front development of capitalized method that more can investment promotion field, and the research and development of industrial technology and innovation are played to positive support effect.

Other will describe in detail in following chapters and sections about concrete technology characteristic of the present invention and non-obvious outstanding effect.

[brief description of the drawings]

Fig. 1 is the first preferred embodiment that the present invention proposes, and is a kind of process flow diagram of constructing method (100) of patent leading indicators;

Fig. 2 is quantity and the industry distribution that Shanghai listed company of exchange analyzes parent;

Fig. 3 is that the industry of Shanghai listed company of exchange analyzing samples distributes;

Fig. 4 is the schematic diagram of the first panel data (141);

Fig. 5 is to be the patent leading indicators (172) of a year the leading phase;

Fig. 6 is to be the patent leading indicators (172) of 2 years the leading phase;

Fig. 7 is to be the patent leading indicators (172) of 3 years the leading phase;

Fig. 8 is to be the patent leading indicators (172) of 4 years the leading phase;

Fig. 9 is the second preferred embodiment that the present invention proposes, and is the process flow diagram of the leading equational constructing method of a kind of patent (500).

Figure 10 A, Figure 10 B, Figure 10 C are in the second preferred embodiment of proposing of the present invention, the generative process of patent core index (542).

Figure 11 is the 3rd preferred embodiment that the present invention proposes, and is a kind of process flow diagram of the method (600) of assessing enterprise investment potentiality.

Figure 12 is the 4th preferred embodiment that the present invention proposes, and is a kind of configuration diagram of the computer system (700) of assessing enterprise investment potentiality.

[embodiment]

The present invention mainly discloses a kind of constructing method (100) and application of patent leading indicators, the wherein ABC of related patent information, patent index (PI), financial index (FI) etc., for having, correlative technology field conventionally knows that the knowledgeable can understand, therefore with explanation hereinafter, no longer do complete description.Meanwhile, graphic with what hereinafter contrasted, only express the signal relevant with feature of the present invention, also do not need according to the complete drafting of physical size, formerly explanation.

Please refer to Fig. 1, for the first preferred embodiment of the present invention's proposition, for a kind of constructing method (100) of patent leading indicators, patent leading indicators (172) are in order to predict the financial information of patent entity (PE), and the information of the financial index (FI) of the leading patent entity of the information of patent leading indicators (172) (PE) has the leading phase of predefined time (L), the constructing method (100) of patent leading indicators comprises:

Step 110: set multiple patent entities (PE) and multiple patent indexes (PI) and financial index (FI) in order to describe each patent entity (PE), each patent index (PI) is obtained by the patent information computing of each patent entity (PE).

Step 120: setting data is collected the phase (121), collection period (121) is made up of time interval (T) and time issue (N), and time issue (N) is for being not less than two integer.

Step 130: collect each patent entity (PE) in collection period (121), the corresponding patent index data (131) of each time interval (T) institute and financial index data (132).

Step 140: by multiple patent index data (131) of multiple patent entities (PE) and multiple financial index data (132) composition the first panel datas (141).

Step 150: the second panel data (151) that the first panel data (141) is formed to normal distribution and criterion score by translation operation program (152).

Step 160: leading phase of setting-up time (L) and the very first time Sequence Operation Theory program (161) based on the leading phase of time (L) is provided, the leading phase of this time (L) comprises at least one time interval (T), the independent variable of very first time Sequence Operation Theory program (161) is the patent index data (131) of a patent index (PI) of the second panel data (151), and the dependent variable of very first time Sequence Operation Theory program (161) is the financial index data (132) of the financial index (FI) of the second panel data (151).

Step 170: set first threshold (171), successively use very first time Sequence Operation Theory program (161) and leading phase of time (L), computing the second panel data (151), from multiple patent indexes (PI), screening draws at least one the patent leading indicators (172) that meets first threshold (171).

In above-mentioned steps 110, the subject of right of patent entity (PE) for having patent right and can making a profit by patent right operation, is preferably the listed company of public publication, but is not limited with listed company; The present embodiment is also applicable to private company, enters as long as can accept the external sources of finance, shares the subject of right of equity and return on equity, all belongs to the scope of application of the present embodiment.Again, the first preferred embodiment for patent, not limiting is granted patent, as long as the patent announced in patent database all can, comprise disclosure of the invention patent, invention granted patent, utility model patent, design patent etc.Meanwhile, the method that the first preferred embodiment proposes is different from the problem of United States Patent (USP) except can effectively solving the information content of China's Mainland patent, in fact more go for all parts of the world district patent.

Aspect patent index (PI), taking the patent of China's Mainland as example, patent index (PI) includes but not limited to that the following can be by the quantitative index of the automatic computing of computing machine, as:

P1: patent sum

P2: disclosure of the invention patent sum

P3: utility model patent sum

P4: design patent sum

P5: invention granted patent sum

P6: average patent life-span of disclosure of the invention patent

P7: average patent life-span of utility model patent

P8: average patent life-span of design patent

P9: average patent life-span of invention granted patent

P10: average the period under review of invention granted patent

P11: current disclosure of the invention patent number

P12: current utility model patent number

P13: current design patent number

P14: current invention granted patent number

P15: current average the period under review of invention granted patent, from the applying date to Granted publication day only

P16: current disclosure of the invention patent IPC classification number sum

P17: current utility model patent IPC classification number sum

P18: current invention granted patent IPC classification number sum

P19: current disclosure of the invention patent IPC classification number average

P20: current utility model patent IPC classification number average

P21: current invention granted patent IPC classification number average

P22: current disclosure of the invention patent specification total page number

P23: current utility model patent instructions total page number

P24: current invention granted patent instructions total page number

P25: the current average number of pages of disclosure of the invention patent specification

P26: the current average number of pages of utility model patent instructions

P27: the current average number of pages of invention granted patent instructions

P28: the claim sum of current disclosure of the invention patent

P29: the claim sum of current utility model patent

P30: the claim sum of current invention granted patent

P31: the claim average of current disclosure of the invention patent

P32: the claim average of current utility model patent

P33: the claim average of current invention granted patent

P34: the exclusive rights sum of current disclosure of the invention patent

P35: the exclusive rights sum of current utility model patent

P36: the exclusive rights sum of current invention granted patent

P37: the exclusive rights average of current disclosure of the invention patent

P38: the exclusive rights average of current utility model patent

P39: the exclusive rights average of current invention granted patent

P40: the accompanying drawing number of current disclosure of the invention patent specification

P41: the accompanying drawing number of current utility model patent instructions

P42: the accompanying drawing number of current invention granted patent instructions

P43: the accompanying drawing average of current disclosure of the invention patent specification

P44: the accompanying drawing average of current utility model patent instructions

P45: the accompanying drawing average of current invention granted patent instructions

In above-mentioned patent index (PI); " current " data operation that refers to a patent index (PI) is limited in certain special time interval (T); for example; if time interval (T) be year that P11 (current disclosure of the invention patent number) represents all invention disclosed patent numbers within interior 1～Dec in certain year; P15 (current average the period under review of invention granted patent, from the applying date to Granted publication day only) represent in certain year average the period under review of the invention granted patent of all bulletins within 1～Dec.If time interval (T) is season, P11 (current disclosure of the invention patent number) represents all invention disclosed patent numbers within interior 3 months of certain season; P15 (current average the period under review of invention granted patent, from the applying date to Granted publication day only) represent in certain season average the period under review of the invention granted patent of all bulletins within 3 months.This analogizes Yu.

Above-mentioned 45 patent indexes (PI), can divide into disclosure of the invention patent class, invention granted patent class, utility model patent class and design patent class again, wherein have in order to the index of describing disclosure of the invention patent:

P2: disclosure of the invention patent sum

P6: average patent life-span of disclosure of the invention patent

P11: current disclosure of the invention patent number

P16: current disclosure of the invention patent IPC classification number sum

P19: current disclosure of the invention patent IPC classification number average

P22: current disclosure of the invention patent specification total page number

P25: the current average number of pages of disclosure of the invention patent specification

P28: the claim sum of current disclosure of the invention patent

P31: the claim average of current disclosure of the invention patent

P34: the exclusive rights sum of current disclosure of the invention patent

P37: the exclusive rights average of current disclosure of the invention patent

P40: the accompanying drawing number of current disclosure of the invention patent specification

P43: the accompanying drawing average of current disclosure of the invention patent specification

Have in order to the index of describing invention granted patent:

P5: invention granted patent sum

P9: average patent life-span of invention granted patent

P10: average the period under review of invention granted patent

P14: current invention granted patent number

P15: current average the period under review of invention granted patent, from the applying date to Granted publication day only

P18: current invention granted patent IPC classification number sum

P21: current invention granted patent IPC classification number average

P24: current invention granted patent instructions total page number

P27: the current average number of pages of invention granted patent instructions

P30: the claim sum of current invention granted patent

P33: the claim average of current invention granted patent

P36: the exclusive rights sum of current invention granted patent

P39: the exclusive rights average of current invention granted patent

P42: the accompanying drawing number of current invention granted patent instructions

P45: the accompanying drawing average of current invention granted patent instructions

Wherein have in order to the index of describing utility model patent

P3: utility model patent sum

P7: average patent life-span of utility model patent

P12: current utility model patent number

P17: current utility model patent IPC classification number sum

P20: current utility model patent IPC classification number average

P23: current utility model patent instructions total page number

P26: the current average number of pages of utility model patent instructions

P29: the claim sum of current utility model patent

P32: the claim average of current utility model patent

P35: the exclusive rights sum of current utility model patent

P38: the exclusive rights average of current utility model patent

P41: the accompanying drawing number of current utility model patent instructions

P44: the accompanying drawing average of current utility model patent instructions

Have in order to the index of describing design patent

P4: design patent sum

P8: average patent life-span of design patent

P13: current design patent number

The patent index (PI) of above-mentioned P1 to P45 is for describing the most sufficient quantitative index of the every speciality of China's Mainland patent, just there is the effect of prediction financial information as for which patent index (PI) wherein? need to be by rigorous analysis and checking, it is core of the present invention place, in follow-up length, can go on to say.

Aspect financial index (FI), the index of expressing performance that refers to used in the present invention, can be debt paying ability index, operation capacity index, profitability guideline, developing ability index and stock price indices etc., the present invention not be limited.Wherein profitability guideline can be net assets income ratio ROE (Rate of Return on Common Shareholder ' s Equity), assets return ROA (Rate of Return on Assets), earning per share EPS (Earnings Per Share), HSBC MTB (Market-to-Book Ratio) etc.

In above-mentioned steps 120, the unit of time interval (T) can be week, the moon, season, half a year or year etc.Setting-up time issue (N) is in order to collect enough sample datas with time interval (T), to set up the use of model and checking.If time issue (N) is set as 5, time interval (T) is set as year, and collection period (121) is 5 years, needs to collect the sample data of 5 years just follow-up; If time issue (N) is set as 6, time interval (T) is set as the moon, and collection period (121) is 6 months, needs to collect the sample data of 6 months just follow-up.What propose due to the present invention is forecast model, that is the data of data prediction current period of first phase before wanting, or with the data of first phase under the data prediction of current period, therefore, time issue (N) was at least required to be for 2 phases, could verify the conspicuousness predicting the outcome.

In above-mentioned steps 130: if collection period (121) is 2008 to 2012 these 5 years, time interval (T) is 1 year, time issue (N) is 5, must collect each patent entity (PE) in 2008 to 2012 these 5 years, the patent index data (131) of all patent indexes (PI) in each time and the financial index data (132) of financial index (FI).

In step 140, so-called panel data, is called again panel data or comprehensive column data, is time series data and the mixing of cross-sectional data, and refers to the data set of M xsect object being observed at time issue (N), a total M × N data set.Taking the present embodiment as example, if there are 375 patent entities (PE), collection period (121) is 5 years, 2008 to 2012, time interval (T) is year, now just form 375 xsect objects being observed, at 45 patent indexes (PI) in 5 times time (2008,2009,2010,2011,2012) and the data set of financial index (FI).

Traditional time series data is for analyzing the connection of single object being observed in the observed reading (independent variable and dependent variable) of multiple times.Traditional cross-sectional data is for analyzing the connection of multiple objects being observed in the observed reading (independent variable and dependent variable) of single time point.These two kinds of data are not all suitable for the method that the present invention proposes, because the present invention has multiple objects being observed and has multiple time points, each time point has again multiple independents variable and dependent variable.And panel data, for analyzing the association of multiple specific xsect objects being observed in the observed reading of multiple time points, due to increasing of observed reading, can increase estimator sampling precision, obtain more consistent estimator and efficiency estimate and obtain more multidate information, therefore the present invention adopts panel data analysis.

In the present embodiment, between step 140 and step 150, can further comprise a normal distribution (Normal Distribution) check program (145), to each patent index data (131) of collecting and each financial index data (132), check the state of its normal distribution.If because data do not present normal distribution, in the process of setting up analytical model, often do not restrain because error is too high and cause model collapse.Therefore, for the data that do not present normal distribution, must impose suitable translation operation, first be converted into the state of normal distribution, then carry out the analysis of independent variable and dependent variable.

Having that normal distribution-test program (145) is conventional is following several: Anderson-Darling check program, Ryan-Joiner check program, Kolmogorov-Smirnov check program etc., or can observe more easily the normal distribution situation that the coefficient of skewness and coefficient of kurtosis can inference data, the present invention does not limit and adopts which kind of check program.For the data that do not meet normal distribution, need to impose processing computing to generate the data of normal distribution, wherein, Box-Cox converse routine is conventional mode, the present invention is not also as limit.It must be emphasized that, if raw data has roughly presented normal distribution substantially, just do not needed to impose again processing computing.

Above-mentioned steps 150 further illustrates as follows, the distribution curve of normal distribution be substantially a center line in expectation value (mean value) and symmetrical, taking standard deviation as unit to both sides extend launch curve.When mean value and the standard deviation of two kinds of data sets all differ widely, even if these two kinds of data sets are all normal distribution, but yardstick gap can be very large, is unfavorable for analyzing relatively.Therefore in the process of setting up model, preferably,, by the data of normal distribution by a translation operation program (152) processing that tries again, data are done to normalization (Normalization) conversion.Wherein the normal translation operation program (152) using is " Z score " computing.Pass through " Z score " data set after computing, its expectation value (mean value) is all 0, standard deviation is all 1.If the data set of all independents variable and dependent variable is all converted into after the data set of normal distribution and criterion score, more easily from wherein excavating association.

First panel data (141) of the present embodiment, in step 150, after translation operation program (152), is converted to second panel data (151) of normal distribution and criterion score.

In step 160, very first time Sequence Operation Theory program (161) is preferably the Granger Causality testing model (Granger Causality Test Model) of monobasic.Granger Causality testing model be 2003 the Clive of Nobel prize in economics winner Granger (Clive W.J.Granger) start, between the economic variable of sequence analysis time leading with fall behind relation.Its basic concepts are, if have two variable X and Y, variable X occurs formerly, after variable Y occurs in, and by after Granger Causality testing model, variable X is set up in checking the impact of conspicuousness on the probability of happening of variable Y, claim variable X to lead over variable Y, or claims the leading indicators that variable X is variable Y.In the economic variable of Granger Causality testing model processing, its independent variable and dependent variable are all seasonal effect in time series variablees, its basic operational model is regression analysis model, but before regretional analysis, first to independent variable or the side-play amount of dependent variable setting a period of time, set leading phase or the phase of backwardness, then inspect to there is the leading phase or fall behind the regression analysis model of phase and join appropriateness, to verify the leading effect of independent variable or to fall behind effect.

The model of Granger Causality testing model is joined the P_value that the conventional F of appropriateness inspection institute obtains, and general P_value<0.1, for model can be accepted; If P_value<0.05, for model good; If P_value<0.005, for model splendid.Simply say, if independent variable is X, dependent variable is Y, and via the P_value<0.05 obtaining after Granger Causality testing model, is illustrated in 95 percent fiducial interval, and independent variable X is Y for dependent variable, has leading effect.It is R square value that another conventional model of Granger Causality testing model is joined appropriateness, and R square value, between 0 to 1, more approaches 1 better, and it is 0 poorer more to approach; Taking 1 as best, represent model perfection, have no error; For the poorest, represent error infinity with 0.

In Granger Causality testing model, do not limit the number that represents independent variable.That is Granger Causality testing model can be analyzed the leading effect of an independent variable to a dependent variable, now claim the Granger Causality testing model of monobasic, as step 160; Also can analyze the leading effect of multiple independents variable to a dependent variable simultaneously, now claim polynary Granger Causality testing model.But we it must be understood that, if there is serious collinearity between multiple independent variable, or the data discrete of independent variable is too high, and polynary Granger Causality testing model often collapses in calculating process.Hereat, in step 160, preferably, it is the Granger Causality testing model computing that first the patent index data (131) of other patent index (PI) is imposed to monobasic to the financial index data (132) of individual other financial index (FI), could be in step 170, from 45 patent indexes (PI), excavate other patent index (PI) that leading effect has conspicuousness, be called patent leading indicators (172), and get rid of other leading effects and do not have the patent index (PI) of conspicuousness, each patent leading indicators (172) of now excavating have conspicuousness to the leading effect of financial index (FI).

In step 170, the setting of first threshold (171) is crucial, and it is too strict that first threshold (171) is set, and possibly cannot excavate the patent leading indicators (172) with conspicuousness; It is too loose that first threshold (171) is set, and the patent index (PI) of conspicuousness deficiency is also mistaken as patent leading indicators (172) and excavates out too much.If very first time Sequence Operation Theory program (161) is taked the Granger Causality testing model of monobasic, as aforementioned, model is joined the P_value that the conventional F of appropriateness inspection institute obtains, general P_value<0.1, can accept for model, reach 90% fiducial interval; If P_value<0.05, for model is good, reaches 95% fiducial interval; If P_value<0.005 is splendid for model, reach 99.5% fiducial interval.Now we can be 0.1 to first set first threshold (171), first preliminary understanding can be excavated how many patent leading indicators (172) and reach 90% fiducial interval, if quantity is few, setting first threshold (171) is 0.1; If quantity is a lot, can reset first threshold (171) is 0.05 or 0.005, can excavate patent leading indicators (172) crucial, that leading effect has splendid conspicuousness.

In the situation that first threshold (171) is set in 0.05, after excavating patent leading indicators (172), if the leading phase of time (L) of setting in step 160 is 1 year, observe the numerical value of its patent leading indicators (172) of each patent entity (PE) in this year, just can be used as patent entity (PE) the predicting the outcome of the financial index numerical value in next year, and can reach 95% fiducial interval.If first threshold (171) is set in 0.005, and the leading phase of time (L) of setting in step 160 is 3 years, after excavating patent leading indicators (172), observe the numerical value of its patent leading indicators (172) of each patent entity (PE) in this year, just can be used as predicting the outcome of the financial index data of patent entity (PE) after 3 years, and can reach 99.5% fiducial interval.

Be patent entity (PE) by the listed company taking Shanghai exchange below, be described in further detail the implementation process of the first preferred embodiment.

Shanghai exchange, by end in 2012, has 951 listed companies, and the parent of patent entity (PE) is 951, and its industry distribution and proportion are as shown in Figure 2.

The present invention is the forecast model of setting up the primacy of patent index (PI) to financial index (FI), therefore must consider subsidiary company's structure of listed company.If the finance combination of subsidiary company calculates in the lump to parent company, the patent index (PI) of subsidiary company also must merge to parent company and calculate in the lump.So 951 listed companies all must its subsidiary company's structure of investigation.Through investigation after, we find, in 951 listed companies, its subsidiary company's quantity mean value is 13.6, median is 9.0.By the difference of subsidiary company's mean value and median, just can learn that the subsidiary company that most of listed company has estimates below 10, but wherein have minority listed company to there is very a large amount of subsidiary companies, therefore draw high mean value.Through investigation, have the listed company of maximum subsidiary companies, it has up to 174 subsidiary companies.

Aspect the financial index (FI) of listed company, we tentatively select net assets income ratio ROE (Rate of Return on Common Shareholder ' s Equity) as representative, carry out follow-up analysis.Aspect patent index (PI), we adopt 45 patent indexes (PI) such as aforesaid P1～P45.

Aspect the selecting of effective sample, we set, and from 2008 to 2012, the financial index (FI) these 5 years all must have data, and end had at least 50 patent sums (comprising the totallings such as disclosure of the invention, invention mandate, utility model, appearance design) in 2012,951 listed companies qualifiedly finally obtain 375 through screening, the sample that is patent entity (PE) is 375, and its industry distribution and proportion are as shown in Figure 3.

375 patent entities (PE) are collected after the data (131) of its 45 patent indexes (PI) of 2008 to 2012 and the data (132) of financial index (FI), form the first panel data (141), to carry out follow-up analysis.

Please refer to Fig. 4, for 375 patent entities of the present invention (PE), form a part of content of the first panel data (141) in the patent index data (131) of 45 patent indexes (PI) in 5 times time (2008,2009,2010,2011,2012) and the financial index data (132) of financial index (FI).

Generate after the first panel data (141), must be to data wherein, no matter be as the data of independent variable or as the data of dependent variable, analyze its data distribution situation, check and whether meet normal distribution.Only meet the data of normal distribution, just easily set up the relational model of independent variable and dependent variable, otherwise most probably because error causes too greatly model collapse.

By Kolmogorov-Smirnov check program, we find in the first panel data (141), the financial index data (132) of financial index (FI) roughly present normal distribution substantially, and the patent index data (131) of each patent index (PI) do not present normal distribution.Therefore, we impose again Box-Cox converse routine to the patent index data (131) of each patent index (PI) and make it roughly present normal distribution.

Now, our the patent index data (131) of the patent index to all normal distribution (PI) and financial index data (132) of financial index (FI), impose again again Z score translation operation program (152), the expectation value (mean value) that makes each data is all 0, and standard deviation is all 1.Now the first panel data (141) is all converted to second panel data (151) of normal distribution and criterion score.

In the second panel data (151), independent variable is 45 patent indexes (PI) of 375 patent entities (PE) in 5 years, dependent variable is the financial index (FI) of 375 patent entities (PE) in 5 years: ROE, and now we use monobasic Granger Causality testing model (Granger Causality Test Model) to check successively the precondition of each independent variable to dependent variable.

In Granger Causality testing model (Granger Causality Test Model), a very important parameter is the setting of leading phase of time (L).In the present embodiment, we distinguish the leading phase of setting-up time (L) is 1 year, 2 years, 3 years, 4 years, observes the precondition of patent index (PI) to financial index (FI).

When set the leading phase of time, (L) was 1 year time, the independent variable of use is 2008,2009,2010,2011 these patent indexes in 4 years (PI); Corresponding dependent variable is 2009,2010,2011,2012 these financial index in 4 years (FI).Wherein, the patent index (PI) of 2008 is the financial index (FI) matching to 2009, the patent index (PI) of 2009 is the financial index (FI) matching to 2010, the patent index (PI) of 2010 is the financial index (FI) matching to 2011, the patent index (PI) of 2011 is the financial index (FI) matching to 2012, so could verify the conspicuousness of patent index (PI) to the leading effect of financial index (FI).Now, in the second panel data (151), there are 375 × 4=1500 data to use.

When set the leading phase of time, (L) was 2 years time, the independent variable of use is 2008,2009,2010 these patent indexes in 3 years (PI); Corresponding dependent variable is respectively 2010,2011,2012 these financial index in 3 years (FI), now, in the second panel data (151), has 375 × 3=1125 data to use.

When set the leading phase of time, (L) was 3 years time, the independent variable of use is 2008,2009 these patent indexes in 2 years (PI); Corresponding dependent variable is respectively 2011,2012 these financial index in 2 years (FI), now, in the second panel data (151), has 375 × 2=750 data to use.

When set the leading phase of time, (L) was 4 years time, the independent variable of use is the patent index (PI) of 2008; Corresponding dependent variable is the financial index (FI) of 2012, now, in the second panel data (151), only has 375 data to use.

We are used the Granger Causality testing model (Granger Causality Test Model) of monobasic very first time Sequence Operation Theory program (161), model is joined appropriateness, and we use F inspection, and model is joined appropriate P_value and adopted three kinds of first thresholds (171):

(1) P_value<0.1, independent variable patent index (PI) has acceptable conspicuousness to the precondition of dependent variable financial index (FI), reaches 90% fiducial interval;

(2) P_value<0.05, independent variable patent index (PI) has good conspicuousness to the precondition of dependent variable financial index (FI), reaches 95% fiducial interval;

(3) P_value<0.005, independent variable patent index (PI) has splendid conspicuousness to the precondition of dependent variable financial index (FI), reaches 99.5% fiducial interval.

The second panel data (151) is passed through after the computing of monobasic Granger Causality testing model, we successfully find wherein really to have some patent index (PI) to have conspicuousness to the primacy of financial index (FI), are called patent leading indicators (172).Fig. 5 to Fig. 8 is respectively the leading phase to be 1 year, 2 years, 3 years, 4 years and to have the patent leading indicators (172) of conspicuousness, and in figure, * represents P-value<0.1; * represents P-value<0.05; * * represents P-value<0.005.

Fig. 5 to Fig. 8 provides a good forecast function, if we relatively pay close attention to 1 year later portfolio performance, the leading phase shown in Fig. 5 of observing is the numerical value change of the patent leading indicators (172) of a year; If we relatively pay close attention to the portfolio performance after 2 years, the leading phase shown in Fig. 6 of observing is the numerical value change of the patent leading indicators (172) of 2 years; If pay close attention to the portfolio performance of 3 years or 4 years, observe respectively the numerical value change of the patent leading indicators (172) shown in Fig. 7 or Fig. 8.

If we are more concerned about wherein most crucial, the patent leading indicators (172) of leading a year, 2 years, 3 years, 4 years simultaneously, find out the patent leading indicators (172) that appear at Fig. 5 to Fig. 8 simultaneously.As beneath 5 patent leading indicators (172), exactly simultaneously leading 1 year, leading 2 years, leading 3 years with within leading 4 years, all there is conspicuousness person.

P12 (current utility model patent number)

P23 (current utility model patent instructions total page number)

P29 (the claim sum of current utility model patent)

P35 (the exclusive rights sum of current utility model patent)

P41 (the accompanying drawing number of current utility model patent instructions)

P12, P23, P29, P35, these five patent leading indicators (172) of P41 all belong to the patent index (PI) of utility class, reason is that the listed company of Shanghai exchange is taking manufacturing industry (industry code C) as maximum, and include in 375 patent entities of effective sample (PE) of the present embodiment also taking manufacturing industry as at most, and manufacturing patent accounted for major part with utility model in the past, therefore just naturally easily showing one's talent in the process of establishing of forecast model of the patent index of utility class (PI).And this result is also revealed another information simultaneously: even if many experts think that the degree of innovation of utility model patent is lower, it is not high to be worth, but for the listed company of Shanghai Stock Exchange, be but to predict that the performance of its financial index (FI) has the forecasting tool of conspicuousness.

Must emphasize at this, when setting up the effective sample of model, when patent entity (PE) changes, the patent leading indicators (172) of finally showing one's talent will change according to sample properties.For example, when the effective sample of setting up model adopts the listed company of Shenzhen exchange, the patent leading indicators (172) of showing one's talent must have part to be different from the patent leading indicators (172) of being excavated for sample with Shanghai listed company of exchange.Namely, because patent leading indicators (172) can change according to sample properties, the method that the present invention's the first preferred embodiment proposes, has more universality, is applicable to various sample colony.We can, taking information industry listed company as sample colony, excavate the patent leading indicators (172) that are applicable to information industry listed company; Also can be sample colony for Bio-pharmaceutical Industry listed company, excavate the patent leading indicators (172) that are applicable to Bio-pharmaceutical Industry listed company; More can, for chemical engineering of materials industry listed company, excavate separately the patent leading indicators (172) that are applicable to chemical engineering of materials industry listed company.

Explanation and checking that the above-mentioned listed company with Shanghai exchange is done for patent entity (PE), that financial index (FI) adopts is rate of return on equity ROE, but we must emphasize, rate of return on equity ROE is only for convenience of description, and the method for the first preferred embodiment is suitable for various existing financial index (FI).

Please refer to Fig. 9, is the second preferred embodiment of the present invention's proposition, is the leading equational constructing method of a kind of patent (500), and this patent leading side formula (501) is in order to predict the financial information of patent entity (PE).Patent leading side formula (501) can generate a leading mark of patent (502), and the financial information of the leading patent entity of the leading mark of patent (502) (PE) has the leading phase of predefined time (L).The step of the leading equational constructing method of this patent (500) comprising:

Step 510: according to the constructing method (100) of the patent leading indicators described in the first preferred embodiment, obtain the second panel data (151) and multiple patent leading indicators (172).

Step 520: based on multiple patent leading indicators (172), form the 3rd panel data (521) from the second panel data (151) screening.

Step 530: the second time series operation program (531) based on the leading phase of time (L) is provided, the independent variable of the second time series operation program (531) is all patent leading indicators (172) of the 3rd panel data (521), and the dependent variable of the second time series operation program (531) is the financial index (FI) of the 3rd panel data (521).

Step 540: set Second Threshold (541), by the second time series operation program (531) and leading phase of time (L), computing the 3rd panel data (521), from multiple patent leading indicators (172), screening draws and meets multiple patent core index (542) of Second Threshold (541) and generate patent leading side formula (501), and patent leading side formula (501) is made up of multiple patent core index (542) and corresponding weight coefficient (543) thereof in fact.

The fundamental purpose of the second preferred embodiment is for the situation that has multiple patent leading indicators (172), multiple patent leading indicators (172) are combined as to a patent leading side formula (501), being used for by this predicting the finance performance of enterprise, will be more quick, convenient.The data value importation patent leading side formula (501) that the patent core index (542) of patent entity (PE) is had, generates the leading mark of patent (502).The higher person of the leading mark of patent (502), represent patent entity (PE) after the leading phase of time (L) institute the numerical value of financial index (FI) is higher accordingly; The lower person of the leading mark of patent (502), represent patent entity (PE) after the leading phase of time (L) institute the numerical value of financial index (FI) is lower accordingly.Because the height of business finance index (FI) is directly expressed the quality of its business performance, the higher person of numerical value of financial index (FI), performance better, more has investment value.Because the leading mark of patent (502) represents patent entity (PE) institute's numerical value of financial index (FI) accordingly after the leading phase of time (L), therefore observe the high or low of the leading mark of patent (502), just can pick out the object with investment potential from patent entity (PE).

In step 530, the second time series operation program (531) is polynary Granger Causality testing model, its independent variable is multiple patent leading indicators (172) that the first preferred embodiment step 170 obtains, object is that multiple patent leading indicators (172) are combined, and excavates the leading effect to financial index (FI) after multiple patent leading indicators (172) combinations.

But now, we it must be understood that another key concept, polynary Granger Causality testing model is not the simple totalling that multiple monobasic Granger Causality testing models generate result, in the time of the multiple patent leading indicators of combination (172), the conspicuousness of the leading effect of its other patent leading indicators (172) to financial index (FI) can change, and even wherein the leading effect of some patent leading indicators (172) can become not remarkable on the contrary.Therefore in step 530, preferably, can further operate dependent variable delete program item by item.That is, first all patent leading indicators (172) are included in to the dependent variable in polynary Granger Causality testing model, observe the P_value of rear each patent leading indicators (172) of inspection, delete the poorest patent leading indicators (172) of even not having a conspicuousness of conspicuousness, then the polynary Granger Causality testing model of reforming, observe again the P_value of rear each patent leading indicators (172) of inspection, delete again the poorest patent leading indicators (172) of even not having a conspicuousness of conspicuousness, repeat this process, finally leave the patent leading indicators (172) that showing property is higher and be called patent core index (542).Now, polynary Granger Causality testing model can be integrated all patent core index (542) and generate patent leading side formula (501), and patent leading side formula (501) is made up of multiple patent core index (542) and corresponding weight coefficient (543) thereof in fact.

We are again taking the patent leading indicators (172) of leading a year shown in Fig. 5 as example explanation, and these patent leading indicators (172) are all the finance performances for predicting patent entity (PE).Wherein there are the patent leading indicators (172) of part can also be used for assessing unitary patent, as the intensity index of assessment unitary patent, as:

P6: average patent life-span of disclosure of the invention patent

P7: average patent life-span of utility model patent

P9: average patent life-span of invention granted patent

P10: average the period under review of invention granted patent

P15: current average the period under review of invention granted patent, from the applying date to Granted publication day only

P38: the exclusive rights average of current utility model patent

If we want patent leading indicators (172) composition patent leading side formulas (501) such as the P6 of leading a year, P7, P9, P10, P15, P38, first, we are by the second panel data (151), 375 patent entities (PE) were from the financial index (FI) of 2008 to 2012: ROE and 6 patent leading indicators (172): P6, P7, P9, P10, P15, P38 pick out, composition the 3rd panel data (521).

We use polynary Granger Causality testing model now, set Second Threshold (541): P_value<0.05, the independent variable of analyzing is for the first time 6 patent leading indicators (172): P6, P7, P9, P10, P15, P38, dependent variable is ROE, and analysis result is as Figure 10 A.Therefrom can find, the leading effect conspicuousness of each patent leading indicators (172) has changed, and that P15 becomes is the poorest, P_value=0.7833.

While analysis for the second time, we reject the poorest P_value P15, and independent variable uses 5 patent leading indicators (172) such as P6, P7, P9, P10, P38, and analysis result as Figure 10 B is.Wherein P9 is for the poorest, P_value=0.4188.

While analysis for the third time, we reject P9 the poorest P_value, independent variable uses 4 patent leading indicators (172) such as P6, P7, P10, P38, analysis result is as Figure 10 C, wherein each patent leading indicators (172) meets Second Threshold (541), its P_value is less than 0.05, and forecast model reaches 95% fiducial interval.4 patent leading indicators (172) shown in Figure 10 C, now we are defined as patent core index (542).

The basic operational model of polynary Granger Causality testing model is multiple regression analysis model, therefore above-mentioned analysis for the third time except excavating patent core index (542), generated a composite equation formula simultaneously, we are referred to as patent leading side formula (501), wherein

Patent leading side formula (501)=w6 × P6+w7 × P7+w10 × P10+w38 × P38

Wherein w6, w7, w10, w38 correspond to patent core index (542): the weight coefficient (543) of P6, P7, P10, P38, represents the susceptibility that its corresponding patent core index (542) is predicted the primacy of financial index (FI).The present embodiment passes through actual operation, w6=0.1236, w7=0.0236, w10=0.0596, w38=0.0247, now the R square value of patent leading side formula (501) reaches 0.9065, and after adjusting, R square value reaches 0.8750, is good model.In the present embodiment, the value of w6 is the highest, is almost 5 times of w38, represents that P6 (average patent life-span of disclosure of the invention patent) is the most responsive to the prediction of financial index (FI).When the numerical value of P6 (disclosure of the invention patent average patent life-span) and P38 (the exclusive rights average of current utility model patent) all only changes the situation of a unit, the variation that P6 (average patent life-span of disclosure of the invention patent) causes financial index (FI) is exactly 5 times of P38 (the exclusive rights average of current utility model patent) variation that financial index (FI) is caused.

Please refer to Figure 11, is the 3rd preferred embodiment of the present invention's proposition, is a kind of method (600) of assessing enterprise investment potentiality, comprises the following steps:

Step 610: the patent information (612) of collecting multiple enterprises (611);

Step 620: a patent leading side formula (501) being obtained by aforementioned the second preferred embodiment is provided, and patent leading side formula (501) is made up of multiple patent core index (542) and corresponding weight coefficient (543) thereof in fact;

Step 630: the patent information (612) based on each enterprise (611) is calculated the data (631) of the corresponding patent core index of each enterprise (542);

Step 640: the data (631) of the patent core index (542) based on each enterprise (611), generate the leading mark of patent (502) of each enterprise (611) by the calculating of patent leading side formula (501);

Step 650: by a sequencer program (651), the leading mark of patent (502) of each enterprise (611) is sorted, ranking results (652) represents the investment potential of enterprise (611).

The higher person of the leading mark of above-mentioned patent (502), represent enterprise (611) after the leading phase of time (L) institute the numerical value of financial index (FI) is higher accordingly; The lower person of the leading mark of patent (502), represent enterprise (611) after the leading phase of time (L) institute the numerical value of financial index (FI) is lower accordingly.Because the numerical value height of enterprise (611) financial index (FI) is directly expressed the quality of its business performance, the higher person of numerical value of financial index (FI), performance better, more has investment value; The lower person of numerical value of financial index (FI), performance is poorer, does not more have investment value.Because the leading mark of patent (502) represents enterprise (611) institute's numerical value of financial index (FI) accordingly after the leading phase of time (L), therefore the height rank of the leading mark of patent (502) being had by enterprise (611), just can be from wherein picking out the object with investment potential.

Please refer to Figure 12, for the 4th preferred embodiment of the present invention's proposition, for a kind of computer system (700) of assessing enterprise investment potentiality, comprise patent information harvester (710), index calculation element (720), the leading mark calculation element of patent (730) and mark collator (740).

Wherein, patent information harvester (710) is collected the patent information (612) of multiple enterprises.

The data (631) of the corresponding patent core index of enterprise (542) are calculated and generated to the patent information (612) of index calculation element (720) based on each enterprise.

The leading mark calculation element of patent (730) is according to the data (631) of the patent core index (542) of each enterprise, the patent leading side formula (501) obtaining by aforementioned the second preferred embodiment, calculate and generate the leading mark of patent (502) of each enterprise, this patent leading side formula (501) is made up of multiple patent core index (542) and corresponding weight coefficient (543) thereof in fact.

Mark collator (740) sorts leading multiple patents mark (502), and ranking results (652) represents the sequence of the investment potential of multiple enterprises.

The method (600) of assessment enterprise investment potentiality proposed by the invention and computer system (700), it is the achievement based on large data, objective computing, rigorous checking, not only contribute to the technical strength development of patent information analysis and utilization, the front development of capitalized method that more can investment promotion field, and the research and development of industrial technology and innovation are played to positive support effect.

More than explanation, should understand and implement for the special personage of correlative technology field.The foregoing is only the present invention's preferred embodiment, not in order to limit the present invention's interest field simultaneously.Any being equal to of completing based on disclosed content, changes or modifies, and all should be included in the covering scope of claims.

Claims (10)

1. the constructing method of patent leading indicators (100), these patent leading indicators (172) are in order to predict the financial information of patent entity (PE), and the information of the financial index (FI) of leading this patent entity (PE) of the information of these patent leading indicators (172) has the leading phase of predefined time (L), the constructing method (100) of these patent leading indicators is characterised in that the following step:

(1) set multiple patent entities (PE) and in order to describe respectively multiple patent indexes (PI) and the financial index (FI) of this patent entity (PE), respectively this patent index (PI) is obtained by the patent information computing of each this patent entity (PE);

(2) setting data is collected the phase (121), and this collection period (121) is made up of time interval (T) and time issue (N), and this time issue (N) is for being not less than two integer;

(3) collect respectively this patent entity (PE) in this collection period (121), the corresponding patent index data (131) of each time interval (T) institute and financial index data (132);

(4) by the plurality of patent index data (131) of the plurality of patent entity (PE) and the plurality of financial index data (132) composition the first panel datas (141);

(5) this first panel data (141) is formed to second panel data (151) of normal distribution and criterion score by translation operation program (152);

(6) set the leading phase of this time (L) and the very first time Sequence Operation Theory program (161) based on the leading phase of this time (L) is provided, the leading phase of this time (L) comprises at least one this time interval (T), the independent variable of this very first time Sequence Operation Theory program (161) is this patent index (PI) of this second panel data (151), and the dependent variable of this very first time Sequence Operation Theory program (161) is this financial index (FI) of this second panel data (151);

(7) set first threshold (171), successively use this very first time Sequence Operation Theory program (161) and leading phase of this time (L), this second panel data (151) of computing, from the plurality of patent index (PI), screening draws at least one these patent leading indicators (172) that meets this first threshold (171).

2. the constructing method of patent leading indicators according to claim 1 (100), wherein the plurality of patent entity (PE) is listed company.

3. the constructing method of patent leading indicators according to claim 1 (100), wherein the plurality of patent index (PI) comprise describe disclosure of the invention patent index, describe invention granted patent index, describe the index of utility model patent, with the index of describing design patent.

4. the constructing method of patent leading indicators according to claim 1 (100), wherein this financial index (FI) is selected by following formed group: debt paying ability index, operation capacity index, profitability guideline, developing ability index and stock price indices, this profitability guideline at least comprises net assets income ratio.

5. the constructing method of patent leading indicators according to claim 1 (100), wherein this time interval (T) is selected by following formed group: week, the moon, season, half a year, Yi Jinian.

6. the constructing method of patent leading indicators according to claim 1 (100), wherein between this step (4) and (5), further comprise normal distribution-test program (145), check any these patent index data (131) and these financial index data (132).

7. the constructing method of patent leading indicators according to claim 1 (100), wherein this very first time Sequence Operation Theory program (161) is monobasic Granger Causality testing model, this threshold value is not more than 0.1.

8. the leading equational constructing method of patent (500), this patent leading side formula (501) is in order to predict the financial information of patent entity (PE), this patent leading side formula (501) generates a leading mark of patent (502), the financial information of leading this patent entity (PE) of the leading mark of this patent (502) has the leading phase of predefined time (L), and the leading equational constructing method of this patent (500) is characterised in that:

(1) obtain the second panel data (151) and multiple patent leading indicators (172), this second panel data (151) and the plurality of patent leading indicators (172) be by claim 1 to 7 wherein the constructing method (100) of the patent leading indicators described in any one obtained;

(2), based on the plurality of patent leading indicators (172), form the 3rd panel data (521) from this second panel data (151) screening;

(3) provide the second time series operation program (531) based on the leading phase of this time (L), the independent variable of this second time series operation program (531) is all these patent leading indicators (172) of the 3rd panel data (521), and the dependent variable of this second time series operation program (531) is this financial index (FI) of the 3rd panel data (521);

(4) set Second Threshold (541), by this second time series operation program (531) and leading phase of this time (L), computing the 3rd panel data (521), from the plurality of patent leading indicators (172), screening draws and meets multiple patent core index (542) of this Second Threshold (541) and generate this patent leading side formula (501), and this patent leading side formula (501) is made up of the plurality of patent core index (542) and corresponding weight coefficient (543) thereof in fact.

9. assess the method (600) of enterprise investment potentiality for one kind, it is characterized in that:

(1) collect the patent information (612) of multiple enterprises (611);

(2) provide a patent leading side formula (501), this patent leading side formula (501) is obtained by the leading equational constructing method of patent claimed in claim 8 (500), and this patent leading side formula (501) is made up of multiple patent core index (542) and corresponding weight coefficient (543) thereof in fact;

(3) patent information based on Ge Gai enterprise (611) (612) is calculated the data (631) of the corresponding patent core index of Ge Gai enterprise (542);

(4) data (631) of the patent core index (542) based on Ge Gai enterprise (611), calculate the leading mark of patent (502) that generates Ge Gai enterprise (611) by this patent leading side formula (501);

(5) by a sequencer program (651), leading the plurality of patent mark (502) is sorted, this ranking results (652) represents the investment potential of the plurality of enterprise (611).

10. assess the computer system (700) of enterprise investment potentiality for one kind, comprise patent information harvester (710), index calculation element (720), the leading mark calculation element of patent (730) and mark collator (740), it is characterized in that:

This patent information harvester (710) is collected the patent information (612) of multiple enterprises;

The patent information (612) of this index calculation element (720) based on Ge Gai enterprise, calculates the data (631) that generate the corresponding patent core index of this enterprise (542);

The leading mark calculation element of this patent (730), according to the data (631) of the patent core index (542) of Ge Gai enterprise, calculate the leading mark of patent (502) that generates Ge Gai enterprise by a patent leading side formula (501), this patent leading side formula (501) is obtained by the constructing method (500) of patent leading side formula claimed in claim 8 (501), and this patent leading side formula (501) is made up of multiple patent core index (542) and corresponding weight coefficient (543) thereof in fact; This mark collator (740) sorts leading the plurality of patent mark (502) and generate ranking results (652), and this ranking results (652) represents the investment potential of the plurality of enterprise.