Extenics Analysis of Regional Innovation Coordination Ability

The study reviews Chinese and international regional innovation capacity and extenics theory, and applies extenics theory to construct a measurement model of regional innovation coordination. It empirically researches 31 Chinese provinces on the coordination of regional innovation in both 2007 and 2015. The research findings indicate that the Industry-University-Research status is not optimistic in China: on the one hand, the coordination abilities of most enterprises are relatively low; and on the other hand, compared with 2007, the proportion of areas with good coordination dropped 10% in 2015. It is thus clear that the effect of adjustments by relevant government departments relating to Industry-University-Research is poor and the coordination of innovation abilities in some provinces are falling rather than rising. Relevant decision-making authorities should take note of this and take more effective measures to promote the coordination of research cooperation and to enhance regional coordination.


INTRODUCTION
The "Global Competitiveness Report 2016-2017" published by the World Economic Forum ranks China in 28 th position, unchanged from the previous year.The Central Committee of the Communist Party of China, the State Council and entities at all government levels are focused on innovation in Chinese enterprises.The nonprofit social group, China Industry-University Research Institute Collaboration Association, was established in November 2007 to enhance independent innovation capabilities and to promote the commercialization, industrialization and internationalisation of innovative products.This Association built a national innovation base of 31 cooperating industrial parks, alliances and enterprises and initiated an Industry-University-Research collaborative innovation platform based on the Internet that strongly supports and facilitates local Industry-University-Research project activities.The China Industry-University Research Institute Collaboration Association plays an important role in promoting coordination across industry, third level and research.The year 2015 marked the 8-years anniversary of the "China Industry-University Research Institute Collaboration Association", yet competitiveness is declining.Innovation is a key factor limiting the growth of overall competitiveness.The ancient Chinese philosopher Confucius emphasized "a good scholar can become an official" while a wealthy person cannot get a good position in society, moreover the ancient politician Buwei Lv abandoned business for politics.This image has deeply influenced generations of Chinese people, with many people entering government positions and many research personnel joining research institutes and universities, impacting on the personnel structure of Chinese enterprises.This has resulted in an imbalance in regional innovative system that, coupled with poor regional coordination ability, has caused a decline of innovation capabilities in China.Regional coordination ability refers to industry, third level and research coordination within the region and the effective cooperation between regional innovation entities thereby enhancing overall competitiveness.Therefore, industry, third level and research coordination is crucial to resolving China's low regional innovation coordination abilities.

LITERATURE REVIEW
There are many studies on innovation capacity from both a macro (national innovation system) and micro (regional innovation systems) perspective.The era of the knowledge base fosters the national innovation system (Chung, 2002).Scholars have emphasized innovation entities, but have not sufficiently focused on building an effective national innovation system (Nelson, 1993;Patel and Pavitt, 1994;Samara, 2012).The national innovation system can be viewed from narrow and broad perspectives.The broad concept covers all direct or indirect subjects related to the innovation process and the narrow is those subjects that directly related to the creation of and the use of knowledge (Lundvall, 1988).In-depth studies of the national innovation system reveal that very considerable innovation benefits can be produced with good innovation investment conditions (Andersson and Mahroum, 2008), which attracted further attention (Archibugi and Coco, 2004;Archibugi, 2009;Mahroum, 2013).Furman (2007) carried out a detailed study on the national innovation system performance indicators and explored why different countries have different innovation capacities from an empirical analysis perspective.Some experts consider regionalization research to be most important and in the context of globalization, the country level is less important than the economy, research and development and innovation activities (Chung, 2002).There is an increasing emphasis on regional innovation research (Braczyk et al, 1996;Cooke, 1997Cooke, , 2000Cooke, , 2001;;Doloreux, 2002;Hsueh & Su, 2016).From a regional economic development perspective, regional innovation is gradually highlighted (Gerstlberger, 2004).From a relatively macro perspective, the core regional innovation system consists of public, private, and public-private parts (Braczyk, 1998) and its main elements include innovative environment, third level institutes, public institutions and enterprises (Buesa, 2006).Theoretical and empirical research shows that the quality of third level academic research institutions, and the concentration, and the impact of foreign investment are the most important factors affecting regional innovation capability (Fritsch and Slavtchev, 2007;Haskel, 2007).Regional resources, behaviour subject ability, and the ability to interact between entities, and the research, education and technology transfer systems are the key factors causing differences in regional innovation (Todtling, 1998;Braczyk, 1998;Cook, 1998;Tsai, & Lei, 2016).The greater the differences among regions, the more attention should be paid to innovation system study (Buesa, 2006).Buesa (2010) used the knowledge production function for analysis, and pointed out that the national and regional environments, and the innovative companies, third level institutes, and research institutions have an important impact on regional innovation capacity.
The literature review revealed some studies on regional innovation capability, but none on regional innovation coordination capability.As outlined in the introduction, we identified regional innovation coordination capability as the key element in resolving China's current low regional innovation capacity; hence, this paper uses extenics theories and methods, which will be introduced in the second part, to study the issues of regional innovation coordination.Current studies in extenics include: Wang (2009) combined extenics theory and genetic algorithm and proposed a so-called "extension genetic algorithm" clustering method; Ye (2009) applied extenics to gasoline engine misfire fault diagnosis; and Chao (2010) applied extenics theory to the maximum power tracking intelligent algorithm of solar radiation and ambient temperature.This paper uses extenics theory to measure the detailed situation of different regional innovation coordination capabilities in China from a quantitative analysis perspective.

BASIC KNOWLEDGE OF EXTENSION MODEL
Motifs are extension model logic cells that include matter, affair and relative elements (Chen, 2006).We can present a complex phenomenon by using its composite element or motif expression.

Contribution of this paper to the literature
• This is the first time the extenics model has been applied to the study of regional innovation coordination capability and thus expands the area of extenics empirical research.
• Using the extenics theoretical analysis model, this paper set up a regional innovation coordination measurement model and researched regional innovation coordination ability levels.The results show that not enough provinces have good regional innovation coordination ability, and most enterprises have low abilities.Although relevant government departments have adjusted their policies on Industry-University-Research cooperation, they are not effective.
• This paper provided a theoretical basis for relevant government departments policy regulations as well as pointing the direction for further strengthening Industry-University-Research cooperation.

Matter-Element
The matter   is the object,   is the feature,   about   's value   forms an ordered triple element  = (  ,   ,   ) that is the basic element of description, and is called the one dimension element;   ,   ,   are three elements of matter-element .Among them,   ,   constitute two-tuples (  ,   ), called the feature element of matter   .Matter   have n features including  1 ,  2 , ⋯ ,   and   about   ( = 1,2, ⋯ , )'s value   ( = 1,2, ⋯ , ) forming an array

Affair-Element
The interaction between objects is called affair, described by the affair-element.The action   , action feature   and   about  1 ,  2 , ⋯   's value  1 ,  2 , ⋯   forming an array

Relative-Element
Any matter, affair, people, information, and knowledge are inextricably linked with other matter, affair, people, information, and knowledge.The relative element is the visualization tool to describe these relationship interactions.The relative O  , n feature value  1 ,  2 , ⋯   and their value  1 ,  2 , ⋯   constitute the dimension array

Motifs and Class Motifs
For convenience, the matter-element, affair-element and relative-element are called motifs, denoted as Among them, () represents a certain object (matter, action, relative),  1 ,  2 , ⋯   indicate  features of object ,  1 ,  2 , ⋯   represent object  about the feature's relevant value.

REGIONAL INNOVATION COORDINATION CAPACITY MODEL CONSTRUCTION BASED ON EXTENICS
Extenics is the extraction of useful knowledge from large accumulations of data, and it is a knowledge discovery process known as a knowledge discovery database (KDD) (Yang, 2005;Yu, 2005;Chen, 2006;Cai, 2008).
and  indicates the evaluation information element,   ⋅ + is in {} that satisfies (  ) ≤ 0, �        � > 0, and it is an index set that indicates the evaluation information element   , take , max ′ , max ′ �.
Let  min = min (2) The detailed calculation steps of the extenics model are described here and in Section 4 are used to construct a measurement model of regional innovation coordination.
(1) List the original data before the change and after the change, and according to the relative requirement to prescribe eigenvalues  1 ,  2 , ⋯   , give the original evaluation information element as shown in Table 1.
(2) Set the correlation function about the evaluation feature   ()( = 1,2, ⋯ ), and according to the original evaluation information element table calculate the correlation function of the evaluation information element   = �  ,   ,   � before  changes and the correlation function of the evaluation information element   ′ = �  ′ ,   ,   ′ � after  changes.Regularizing the number of associations, gives Table 2.
Among them, the calculation of the correlation function can be divided into two situations.
The first situation is the elementary correlation function whose optimal point is in the middle interval.
Set  0 = ⟨, ⟩,  = ⟨, ⟩, and  0 ∈ , denote the common point of  0 and  as   (  is empty if there is no common point), and for arbitrary  ≠   , let Here () is the maximum value elementary correlation function from the middle of  0 and  on , among them The interval of location relation comprised by ,  0 and , is called the distance between  and interval  0 .

𝐷𝐷(𝑥𝑥, 𝑋𝑋
The location relation between  and nested interval comprised by  0 and , is called the location value of  and nested interval  0 , .
The second situation is the elementary correlation function whose optimal point is not in the middle interval.
When applying elementary functions to solve practical problems, it should be noted that to determine if the optimal value is in the middle, left or right point of the interval must be based on the actual problem, and different types of elementary functions formula are used to calculate this.
The regular formula of the number of association (3) Calculate the Correlation Difference Calculate List the table of correlation difference.Delete the raw 0 value and delete the 0 value column.Relist into a new standardized correlation table .(4) Comprehensive Correlation Degree Calculation The comprehensive correlation degree before  changes: The comprehensive correlation degree after  changes: Comprehensive correlation degree: (5) Compare the comprehensive correlation degree before and after the changes, and according to (  ) ≤ 0 ((  ) ≥ 0) and �        � > 0 (�        � < 0) determine , max 2) , max

ANALYSIS OF REGIONAL INNOVATION COORDINATION CAPACITY BASED ON EXTENICS
We examined research from research and development institutions, enterprises, and third level institutions and calculated their respective external funding to measure their regional innovation capacity.We selected 31 Chinese provinces ( 1 ,  2 , ⋯  31 ) as the research object.Among them, the corresponding research and development institution ( 1 ), industry enterprise ( 2 ) and third level institution ( 3 ) data respectively indicates the external fund sources.As mentioned earlier, the China Industry-University Research Institute Collaboration Association was established in November 2007.Therefore, we chose 2007 and 2015 data for research, to determine the role of the China Industry-University Research Institute Collaboration Association on the development of China's regional innovation.Relative data were obtained from the China Statistical Yearbook (2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016) and the China Statistical Yearbook on Science andTechnology (2008-2016).According to the calculation procedure described in the previous section, we procedure performed it here as follows: (1) Establish the correlation function, and calculate the number of associations before and after the change.
It should be noted, to select the satisfied value ( 01 ) we used our literature research to select the middle value sample from the 31 provinces.For example, in 2007 the provinces ranked in sixteenth position (mid-point) for each indicator d were Tian Jin (186,815),Tian Jin (70,393),and Chongqing (119,415); however, in 2015 they were Heilongjiang (229,667),Jiangxi (109,657),and Jilin (148,084).We selected the corresponding maximum value of each index in our 31 provinces as the satisfied value.
We analysed the elementary correlation function to calculate the number of associations and found they all had the characteristic of the larger the better.Therefore, we had to choose the right distance formula to calculate which could be referred to formulas (1), (3), and (5).

⋯ 𝑣𝑣 2𝑚𝑚
(2) ′′2 (2) (2) Calculation of Association Difference To calculate the association difference according to formula (7), we found that the row or line of association difference 0 does not exist and the corresponding calculation steps do not change (the specific calculation process is omitted).
(3) Calculation of Comprehensive Correlative Degree We used the coefficient of variance method to determine the comprehensive correlative degree according to (8), and (9), with   indicating weight.The benefit of this approach is that objective data determines the weight of indexes, thus avoiding errors caused by subjective reasons.Through the calculation and normalization of the calculation results, we determined the weights of the three indicators as: 0.26, 0.48, and 0.26.Of course, the calculation may be erroneous if only data and not their weights are considered.We must ascertain whether the weight established is reasonable.It is reasonable that the weight of research and development institutions and third level institutions are the same because they all act as enterprise innovation source providers.From international experiences, the former has technology innovation abilities while the latter focuses on basic research.However, in China, research and development institution functions are not perfect, and third level institutions have relatively strong scientific research abilities hence their functions are of equal importance in this context.Enterprise weight reached 0.48, higher than that of research and development institutions and third level institutions.Enterprises are the core of regional innovation and should occupy a relatively dominant position when measuring innovation coordination capacity; hence, their weighting is also reasonable.We thus find that the weight calculated by the coefficient of variation is basically consistent with the actual situation in China, and the specific data is shown in Table 6.
Table 6 shows that only 14 (45.16%)provinces and cities had relatively good regional innovation ability in 2007 (indicated by a positive value), meaning over half of the provinces and cities had poor coordination innovation ability.In 2015, the innovation coordination ability in three areas became negative, and the ratio of good innovation coordination ability decreased to 35.48%.The changed region is shown in Table 7.It is clear that the cooperation among industry, third level and research is not enough and that the regulation of relevant departments did not resolve the issue resulting in the decrease of regional innovation coordination ability.
(4) Obtaining Qualitative Change Knowledge According to the calculation presented in point (3), the corresponding original data and interval of qualitative change is listed before and after the change in Table 8.
Table 8 shows that after the implementation of policies to promote Industry-University-Research in China, the coordination ability of Tianjin (2), Henan (18), and Hebei ( 19) provinces all decreased.Although Table 7 shows that the regional innovation coordination ability improved in parts of China, this did not result in qualitative change.
Our empirical study showed that although China attaches great importance on innovation and established a "China Industry-University Research Institute Collaboration Association", from a practical point of view, after one year of development the innovation coordination abilities of regions are non-decreasing rather than increasing.It is clear that the improvement of regional innovation coordination ability is a complex process and that the relevant departments need to strengthen Industry-University-Research cooperation reflecting the actual situation in China.

CONCLUSION
Using the extenics theoretical analysis model, this paper set up a regional innovation coordination measurement model and researched regional innovation coordination ability levels in 2007 and 2015.Our findings are not optimistic for the Industry-University-Research cooperation status quo.Not enough provinces have good regional innovation coordination ability, and most enterprises have low abilities.Although relevant government departments have adjusted their policies on Industry-University-Research cooperation, they are not effective.In 2015, the coordination ability of three provinces dropped from good to low in comparison to 2007.This is the first time the extenics model has been applied to the study of regional innovation coordination capability and thus expands the area of extenics empirical research.Moreover, this paper provided a theoretical basis for relevant government departments policy regulations as well as pointing the direction for further strengthening Industry-University-Research cooperation.This paper has some shortcomings.Because of time and space limitations, only the coordination of regional innovation is examined, and other regional innovation capacities are not included.The authors will further study regional innovation capacities and make efforts to form a comprehensive research system.

Table 1 .
Original evaluation information

Table 2 .
Regulate the number of associations

Table 3 .
Qualitative change in the domain corresponding to the evaluation of information

Table 4 .
Qualitative change in the domain correspond to the information element

Table 5 .
Regulate the number of associations

Table 6 .
Qualitative change in the domain corresponding to the evaluation of information

Table 7 .
Qualitative change in the domain correspond to the information element �  () � �        () �

Table 8 .
Qualitative change in the domain corresponding to the original data table and the corresponding interval