The discovery and surveying of technology trends is fundamental and critical to the survival of companies. As technologies constantly evolve and product life cycles become shorter (Ali, Krapfel, & LaBahn, 1995), the need for effective information and knowledge management has become a crucial issue in the highly competitive economic environment (D. Z. Chen, Chang, Huang, & Fu, 2005). Efforts have been made to study the dynamic nature of technology and the innovation process, and these have focused on the examination of technology structures, innovation histories, and the changes of technology dominators in a given field (S. H. Chen, Huang, & Chen, 2012; M. H. Huang, Chen, Lin, & Chen, 2014; Small, 2006).
Patent analysis methods can be used to capture the evolution of technology topics, as patents represent the innovation output of a company and can be viewed as a means of evaluating the technology trends in an industry (Breitzman & Mogee, 2002; Madani & Weber, 2016). Various conceptual models have been proposed for exploring how “technology communities” evolve over time. Previous studies have taken varying approaches to identify the trends in certain technological fields (Bruck, Réthy, Szente, Tobochnik, & Érdi, 2016; Lee, Yoon, Lee, & Park, 2009), and such methods have also been applied to the scientific literature (M. H. Huang & Chang 2015; Zhao & Strotmann, 2014). Essentially, a technology community is composed of a collection of patents and is where inventors share passions for particular problems, apparatuses, fabrications, or methods. Patents are then interacted with to expand the expertise field, where successful innovations with novelty, enhancements, additional functions, or even breakthroughs are disclosed.
Generally, the generation of invention processes is propelled by a minority of contributors, called “technology dominators”, who understand the basic technologies of a certain field and aim to drive innovation. Technology dominators undoubtedly define the global structure of technology, participate in the longitudinal tracking of innovation, and dominate the crucial trends. In a constantly changing environment, technology dominators can be substituted by new ones, and the cycle of prosperity and decline of dominators does not cease. Understanding such an evolving history and what has changed can lead to a clearer comprehension of innovation development and the possession of a major competitive advantage.
Sensing the need for tracing technology evolvement, a novel method for detecting the evolving trends is proposed in this study. Time series analysis and social network analysis (SNA) are combined with patentometrics to uncover the hidden structure of transition, which can be gained from observing the dynamics of dominators and communities. The solar cell patents between 1997 and 2011 in the United States Patent and Trademark Office (USPTO) database are taken as illustrative samples. Specifically, in the SNA perspective, patent documents are conceived as vertices and text similarities between patent document pairs are conceived as links. By using a clustering algorithm, technology communities were grouped by similar content patents within a full-time data set. This enabled the thematic research topics of each community to be recognized over a series of nonoverlapping time periods with the assistance of high-frequency and unique terms to demonstrate the evolution of technology.
It has been suggested that major patent contributions come from only a few participants during technological evolution (Pilkington, Lee, Chan, & Ramakrishna, 2009). We introduce a quantifying scheme that can be applied at various levels (country, assignee, or inventor) for identifying the technology dominators over time. In this study, our analysis was concentrated at the patent assignee level. The technological developments of each community was divided into a series of separate time periods, and the technology dominators corresponding to each period were then identified by their higher number of patents and citations. Thus, technology dominator status was determined according to the evolving technologies over time. They are the pilots of global developments in innovation, commercialization, and integrated application within a field.
Furthermore, this study not only explored the evolving history of technology dominators but also sought to determine the influence of patent characteristics on technology dominators’ states of appearing, being stable, or exiting, which were the main transition patterns analyzed in this study. Because business markets are competitive, assuming proper control of patent characteristics related to transition patterns is desired by diverse types of stakeholders to potentially facilitate monitoring and predicting rivals’ transition patterns at early stages and to provide valuable assistance in adapting strategies (Aharonson & Schilling, 2016).
A series of statistical testing procedures was used to conduct a preliminary study to gain insight into the significance and directionality of technology dominators’ transition characteristics. The patent characteristics of technology dominators for each snapshot were extracted, which included the technology cycle time (TCT), science linkage, pendency period, originality index, and endogeneity index. Such patent indicators have fixed values and can be calculated as soon as a patent is issued. To enable organizations to examine the possible reasons for characteristic variations in technology dominators’ transitions, this study analyzed transition patterns’ directionality according to the rates of change in patent characteristics.
By using the methodology proposed in this study, critical insights into the major trends in a technological field can be revealed. Such information not only would be useful to the technology management and planning of companies but can also be of reference to policy makers in the creation and development of relevant policies.
To explore the trend of technology dominators and technology-related concerns, as well as their patent characteristics under different transition patterns, this study was conducted through the following steps:
- Retrieved a certain area’s patent documents and sorted them by issue date;
- Built a patent network by using text similarities among retrieved patents;
- Clustered the patent network into technology communities;
- Excluded weak communities;
- Explored the technology topics and technology dominators of all nonoverlapping snapshots and identified their patent characteristics;
- Evaluated the mean value of patents to identify the directivity of indicators with respect to the technology dominators’ transition patterns.
2.1 Step 1. Retrieve Patent Documents
Because of the increasing willingness to invest in and protect research-and-development (R&D) output (Liu et al., 2011), patent analysis that enables companies to monitor technology developments and evaluate competitors’ positions in the market was introduced for this study. In Step 1, the current study took advantage of assignee analyses of solar cell fields and used this area as an example. Confronted with the challenge of global climate change and energy shortages, solar energy is considered an acceptable renewable energy source.
The search rules of patent collection were adopted from the United States Patent Classification (USPC) system. A USPC list of the eligible classifications for solar cell fields was provided by the Pilot Program from the USPTO (2009) and contained parts of photovoltaic (136/243–265) and solar cells (438/57, 82, 84–86, 90, 93, 94, 96, 97) in current USPC classes.
In total, this study retrieved 3,820 issued patents from the USPTO Patent Full-text and Image Database dated from 1997 to 2011. The attribute data (i.e., application date, issue date, title, abstract, and USPC class) and the relational data (i.e., patent or literature citation) for each of the retrieved patents were then downloaded.
2.2 Step 2. Build a Patent Network Using Text Similarities
This study utilized text-based document analysis to measure similarities among documents. Documents with a significant impact were identified through the following steps: collecting the titles and abstracts of a patent set, extracting the term vectors of each document through lower case conversion, removing numbers and punctuation, singularizing and synonymizing the text, eliminating stop words, and forming l-level term hierarchies that present a general–specific lexical level from top to bottom. A self-programming R toolkit was used to analyze patents (including the igraph, tm, RWeka, openNLP, stringr, gdata, proxy, Snowball, and WordNet packages).
In accordance with S. H. Chen, Huang, & Chen (2012), the term length was truncated at the l-level so that the term frequency of the l-level was scarcely greater than (l + 1) level or higher. A term–document matrix was generated using a count-based method, and the patent similarity was then measured using Salton’s cosine. Furthermore, the Salton’s cosine average and standard deviation were used to build a relative threshold to retain strong similarities (S. H. Chen, Huang, Chen, & Lin, 2012).
2.3 Step 3. Clustering the Patent Network into Technology Communities
To construct a patent network and identify the corresponding technical communities from the vertices and ties information of a given full-time data set, the Girvan– Newman (GN) algorithm (Newman, 2004) was adopted. The GN algorithm was chosen because human judgment is not required in setting the number of communities and because it can be used to form a community structure in a given weighted network with high sensitivity and reliability. The GN algorithm calculated the betweennesses of ties, obtained the weighted betweenness by identifying the similarity of the corresponding tie, and then removed the tie with the highest weighted betweenness.
2.4 Step 4. Excluding Weak Communities
In Step 4, the GN algorithm recalculated the weighted betweenness of all the ties on the remaining network that had been affected by the removal of the tie in the previous step, and then removed the tie with the highest weighted betweenness again. The algorithm’s steps for community detection are summarized as follows:
- Calculating the betweenness of all existing ties in the network.
- Removing the tie with the highest weighted betweenness.
- Recalculating the betweenness of all ties affected by the removal.
- Repeating Steps 2 and 3 until no ties remain.
By recalculating the betweennesses after the removal of each tie, it was ensured that at least one of the remaining ties between two communities always had a high value. After the iterations stopped, each split network was compared with the others by means of modularity, an objective value. The highest modularity indicated that a split network had the optimal structure with minimal between-cluster ties and maximal in-cluster ties (Newman, 2004).
2.5 Step 5. Exploring Technology Topics and Technology Dominators and Identifying Their Patent Characteristics
Prior to Step 5, patents have been clustered as several communities and the technology dominators have been identified. The technology dominators had above-average numbers of granted patents and citations. Deriving from Ernst (1999) and Pilkington, Lee, Chan., & Ramakrishna (2009), this scheme enables researchers to identify productive technology dominators with highly cited patents. It also reveals that knowledge is not equally distributed among populations and can be concentrated in a few individuals.
To explain more clearly the transition patterns of technology dominators, this study used the following pattern characteristics: application date, issue date, backward citation, patent reference, nonpatent reference (NPR), and USPC. In the current research, the criteria for selecting characteristics were currency and time-invariant properties. For currency, the characteristics required being available quickly after a patent was issued. From the perspective of time invariance, the characteristics must be aptotic after the issue date of a patent. In addition, the statistical testing was restricted to time-invariant structures; thus, time-variant characteristics that required a trigger time or long range (e.g., generality, patent age, and prosecution length) were excluded. In this study, five characteristics were chosen and are described as follows.
- TCT: This indicator is the median age of the patents cited on the front page of the patent document and calculated from the issue date of the cited patent to the date of each citing patent (Narin, 1994). The TCT indicates the connection between an invention and the most related prior knowledge (Kayal, 1999). It also works on the assumption that the smaller the age of the cited patents is, the more quickly one generation of inventions replaces another (Park, Shin, & Park, 2006). In other words, a short TCT indicates fast technological progress.
- Science linkage: The frequency of the patent references to science articles (i.e., NPRs) indicates the linkage between technologies and academic research results of an analyzed unit (Boyack & Klavans, 2008; Fan, Liu, & Zhu 2017). The NPRs are extracted and counted after the references of all the granted patents are collected (Y. H. Huang, 2009; Lo, 2010). Science linkage has validity for evaluating the effect of science on technology (Narin, 1991, 2000).
- Pendency period: This indicator refers to the period between the filing of a patent application and the issuance of the patent. A higher pending period value indicates a longer examination process and results in the later granting of a patent. As Haupt, Kloyer, & Lange (2007) suggested, a wide range of claims and new innovations during the appearing stage leads to a longer examination process.
- Originality index: Patent originality refers to the wide variety of technological fields on which a patent relies. Developed by Trajtenberg, Jaffe, & Henderson (1997), the originality index is grounded in the calculation of backward citations and indicates the extent of an invention’s reach (Caillaud & Ménière, 2014). Because the originality index is based on inventions’ technology categories, it uses a USPC category histogram that represents the citing patents of an analyzed unit and was calculated using the Herfindahl–Hirschman index (Leydesdorff & Rafols, 2011).
- Endogeneity index: This index examines cited–citing overlap. Because a cited patent may also be a citing patent in a set of patents, the extent to which cited patents overlap with citing patents may be a relevant indicator of their potential growth of technology (Upham & Small, 2010). According to Upham & Small (2010), the level of endogeneity is generally higher for technologies in the emerging state.
2.6 Step 6. Identifying the Directivity of Indicators with Respect to Technology Dominators’ Transition Patterns
After the technology dominators’ characteristics were calculated and normalized, the outcome values were changed into the change rates between two consecutive snapshots. By evaluating these consecutive snapshots, technology dominators’ transition patterns were identified. Transition patterns in this study are represented by appearing, stable, and exiting; these three transition pattern types indicate when technology dominators are appearing, stable, and exiting in their community, respectively (Table 1).
Transition Pattern Types.
|Appearing||A technology dominator emerges from the latter snapshot.|
|Stable||A technology dominator exists in two consecutive snapshots.|
|Exiting||An existing technology dominator disappears in the latter snapshot.|
To examine the change rate of characteristics and identify the transition pattern types, one- and two-tailed t-tests were conducted to calculate the mean values of the various patent characteristics.
First, a one-sample one-tailed t-test was used to determine the directionality of the mean value, which could be nonsignificantly more than zero, less than zero, or equal to zero. Second, when multiple characteristics’ transitions had similar directionality, two-sample one-tailed t-tests were conducted to determine which one’s mean was higher. Levene’s test was then used to examine variance homogeneity. If the p-value was <0.05, it was concluded that the variances were unequal and a t-test based on unequal variance was used for mutual comparison; otherwise, a t-test based on equal variance was used.
3 Results and Discussion
3.1 Technology Topics and Technology Dominator Evolution at the Assignee Level
To understand technology topics and technology dominator evolution in the solar cell field, this study analyzed patents by conducting the aforementioned six-step methodology. All patents with text similarity were collected to form a patent network, and 63 communities were identified using the GN algorithm. Furthermore, Rosin’s (2001) unimodal thresholding was utilized to set a threshold value to filter the dominant communities, and the value 0.041 was identified. Only eight communities exceeded the threshold of 0.041. To identify the transitions in the subcommunities and dominators, each community was then separated into 3-year isolated snapshots. The division of time windows is derived from the TCT of the technology field, which reflects the changing pace of the technology. According to Hirschey & Richardson (2001), the TCT of emerging technologies is 4 years or less. Thus, in consideration of the fast-evolving pace of solar cells, a 3-year window was chosen to reflect the rapid development. The descriptive names and key assignees were then assigned as follows:
Community 1: Organic Electric Devices
Organic solar cells have lower cost and more favorable optical transmission than do other types of solar cells, but their mass production is difficult even with the continual introduction of new fabrication methods. Since 2003, inventors have focused on using organic materials in the production of electric devices such as electric transistors and organic thin film solar cells. Organic semiconductor structures, such as core–shell structures and nanostructures, were then presented. Based on the new structures, enhanced performance could be obtained for electric devices. Among these technology topics, Cambridge Display Technology Limited, Advanced Micro Devices, and Koninklijke Philips Electronics were popular only at the early stage, before fragmentation of the technology among rival competitors began (Figure 1).
Community 2: Organic Solar Cell Modules
Community 2 was associated with the modules of organic solar cells (Figure 2). Using organic materials such as waterproof polymeric resin to accomplish distinct goals is a common means of structural enhancement. Various nanostructures were then designed because they could improve the performance of the modules. Recently, inventors have focused on dye-sensitized solar cell modules, which exhibit favorable optical transparency, flexibility, and conversion rates as well as lower environmental impact and cost. The frequency of assignee variation is not severe in this field. Canon, the Trustees of Princeton University Konarka Technology, Nanosolar, and Nanosys continually dominated this technology over the three periods. In recent years, Semiconductor Energy Laboratory has reestablished activity, and many other newcomers have emerged.
Community 3: Compound Solar Cells
Compound solar cells have the advantages of high conversion rate, high stability, and less pollution. Initially, the properties of semiconductor thin films that were mainly focused on comprised enhancing the conversion rate and connectivity. Inventors then adjusted the material compositions to exhibit various features. Additionally, nanostructures were also applied to increase the efficiency and maintain the stability of compound solar cells. Key assignees were replaced when technology topics evolved. Nanosolar and Solopower dominated in the developing material topics in the most recent two periods. Recently, Azur Space Solar Power GmbH, Konarka Technology, Korea Advanced Institute of Science and Technology, and LG Electronics have emerged as major contributors (Figure 3).
Community 4: Structural Characteristics Analysis of Silicon Solar Cells
Community 4 was related to the analysis of the structural characteristics of silicon solar cells. In the beginning, the structures of the silicon solar cells were investigated, with the characteristics of each layer adjusted to improve efficiency. Inventors then expended a long-term effort to increase production efficiency. Several fabrication processes were improved using different material compositions by enhancing the contact between different layers or by designing different fabrication methods. Recently, nanomaterials have been applied to form a structure that enhances the stability of cells. Semiconductor Energy Laboratory and Sharp were popular for the first four periods, but these were later replaced by several newcomers such as Applied Materials, the Atomic Energy Council, and the Industrial Technology Research Institute (Figure 4).
Community 5: Fabrication Methods and Characteristics of Photoelectric Conversion Devices
For a long period, research has aimed at enhancing the conversion efficiency of photoelectric conversion elements. As the structures and materials of photoelectric conversion elements changed, the corresponding fabrication methods were also changed and improved. Recently, stacked photovoltaic elements have been proposed for shrinking the space required and implementing the concept of three-dimensionality into practice. Additionally, numerous fabrication methods and corresponding characteristics of such elements have been patented. In this context, Canon has been continually active, Asahi Glass has been popular for the last two periods, and Kaneka has recently reemerged (Figure 5).
Community 6: Modules of Silicon Solar Cells
In 1999, different arrangements of modules were enclosed for efficient solar batteries. For enabling mass production, several fabrication methods and rapid installation methods were proposed. For enhancing photoelectric conversion, different solar battery structures that could limit the path and position of electrons were proposed. In recent years, materials such as aluminum, germanium, and boron have also been applied to solar batteries. The concept of three-dimensional solar cells was devised to increase the rates of light trapping and conversion. This technology has seldom been dominated by any company, except for Canon in the early period. The most recent leader was Sumitomo Electric Industries (Figure 6).
Community 7: Carbon Nanotubes
A conjugated system is a unique electric characteristic of carbon nanotubes. Recently, inventors have combined different metals with carbon nanotubes to yield composite materials that have high stiffness, are heat resistant, and can be used to improve the efficiency of solar cells. This is a technology of fast assignee turnover, possibly because of its immaturity and highly competitive nature. Assignees have seldom dominated for longer than two periods, with Samsung being the exception. Compared with other communities, this technology has the most research institutions involved (Figure 7).
Community 8: Silicon Solar Cell Components
Community 8 mainly refers to the components of silicon solar cells. Before 1999, there were only a few inventions that utilized indium phosphate tunnel junctions. Since then, many fabrication procedures have been devised to facilitate the production of solar cells, such as screen printing, masking, and photolithography. Nanostructures and flexible, rigid, and (non)planar surfaces were applied to solar cells to improve efficiency. Additionally, front electrodes have also been improved to reduce visual light reflection, increase conductivity, and lower production costs. This technology has seldom been continually dominated by any company, and the key assignees have changed markedly over time. In recent years, a number of new assignees have risen, and this technology has been controlled by seven assignees (Figure 8).
According to the results, a complete picture of the solar cell family can be described through the combination of communities. Various types of well-known solar cell components are included, such as organic solar cells (Communities 1 and 2), compound-based solar cells (Community 3), silicon-based solar cells (Communities 4, 5, 6, and 8), and nanotechnology solar cells (Community 7).
Based on the evolving technology topics and key assignees, the shift in the solar cell family of several assignees is summarized in Table 2, thus providing a historical overview. For example, during 1997–2005, Fujifilm was at first a key assignee in the silicon-based field, and then it focused its research on organic solar cells and continued with the new technology topic of nanotechnology solar cells.
Shift of Solar Cell Family of Key Assignees.
3.2 Transition Characteristics at Assignee Level
The status of key assignees may be that of appearing, exiting, or being stable in a specific technological evolution and can be discovered by understanding the characteristics of the assignees. Thus, the chosen characteristics of all assignees were utilized to identify transition patterns according to their existence or nonexistence. Consequently, 60 stable, 107 appearing, and 81 exiting patterns were identified.
A one-sample t-test was then used to evaluate the mean values of different transition patterns’ characteristics. For each characteristic, if there were two consistent results in the one-sample t-tests, then two-sample t-tests were used to determine whether the means of groups were equal. The test results showed that only the characteristics of science linkage and originality index required two-sample t-tests (Table 3), and the results of both two-sample t-tests were significant (Table 4). The transition patterns of the characteristics are shown in Figure 9. A double plus or double minus sign indicates that the mean of a characteristic is more or less than zero, respectively. Stable assignees sought to reduce time between the granting of citing patents and the granting of cited patents. Appearing assignees strove to obtain the most relevant and recent knowledge to maintain competitive advantages and react quickly to competitors’ innovations. However, the opposite tendencies were exhibited in the TCT of the exiting assignees, possibly because these assignees were behind the latest core techniques.
One-Sample t-Test of Patent characteristics for Transition Patterns.
|One-sample one-tailed t-test||If two or more directions of one-sample t-tests are similar|
|Technology cycle time||0.841||1.00||0.001***||x|
|Science linkage||0.995||0.001***||1.00||Stable vs. Exiting|
|Originality index||0.974||0.000***||1.00||Stable vs. Exiting|
Two-Sample t-Test of Patent Characteristics for Transition Patterns.
|Characteristics A||Levene’s test||Two-sample one tailed t-test|
The appearing assignees increased their science knowledge citation rate. When novel innovations are being developed, science is the knowledge source. The behavior of citing literature thus increases at the appearing phase. At this phase, there is plentiful novel knowledge published in the scientific literature. By contrast, the science linkage of exiting and stable assignees showed a decreasing trend. This likely caused assignees to withdraw from competition gradually because of inability to gain inspiration from the scientific literature. However, because the stable assignees were required to source knowledge from public science to maintain innovative activity, the science linkage level of the stable assignees was higher than that of the exiting ones.
The examination processes of patents took longer for appearing assignees. This may be because new applicants tended to draft their claims as broadly as possible to reduce the chances of future applicants applying for similar patents. Additionally, the examination lasted longer because the examiners still lacked specific experience concerning the new technologies of appearing assignees. By contrast, the pendency period of exiting assignees was reversed, with assignees eliminated because subsequent inventions were functional enhancements, supplementary to existing ones, or modifications rather than completely novel. It was beyond expectation that the pendency period of the stable assignees was not significantly different from zero. A longer pendency period was expected, because the assignees were compared with higher technological standards to maintain their positions.
Novel patents are innovative and based on broader technological roots. The behavior of citing sources thus increased at the appearing stage because at this stage innovations were created through the synthesis of knowledge across a wide variety of disciplines. The originality index of exiting and stable assignees was the opposite. This may be why assignees withdraw from competition gradually, because their inventions come from relatively monopolistic technologies. However, we found that the originality index level of the stable assignees was higher than that of the exiting ones. This is probably because the stable assignees still required some sources for diverse sets of technologies to maintain novelty.
The endogeneity index showed the capability for knowledge imitation as an increasing trend of the cited–citing density among assignees’ own grant patents. This implies that the appearing assignees successfully accumulated the capability to benefit from past innovations. The endogeneity index of exiting assignees was counter to that of appearing ones. The cited–citing density among the patents of the exiting assignees decreased because sustaining innovation does not continue to create new markets.
4 Conclusions and Suggestions for Future Work
When R&D investment must be increased to maintain a competitive advantage, stakeholders require a global perspective of the technology research topics and technology dominators. This study conducted a patentometric analysis to obtain a general assessment of the field of solar cells. According to the research findings, organic solar cells (Communities 1 and 2) constitute one of the newer and most relevant research topics receiving the most attention from inventors because they reduce the cost of solar power by utilizing inexpensive organic and polymer materials rather than the expensive crystalline silicon used in most other solar cells. In these communities, DuPont, Semiconductor Energy Laboratory, Trustees of Princeton University, and Canon consistently lead the world, though Dai Nippon Printing, Infineon Technology, Samsung, and others have begun to catch up recently. The compound solar cell field (Community 3) was founded by a group that has been losing momentum in recent periods, led by Matsushita Battery, while new generations with many assignees, such as Nanosolar, Solopower, and Azur Space Solar Power are appearing. Silicon solar cells (Communities 4, 5, 6, and 8) constitute an aging topic, probably because it is a mature area and has been implemented in commercial manufacturing. The number of patents decreased continually until 2008, after which the trend exhibits an increase. In this group, Semiconductor Energy Laboratory and Canon are the leaders, and Applied Materials, Atomic Energy Council, Industrial Technology Research Institute, and others have emerged more recently. Finally, nanotechnology solar cells (Community 7) have gradually become an accepted alternative technology, dispersing nanomaterials or nanostructures into the photoactive layer to create devices that are more efficient. In this group, Samsung is consistently the leader, and Hon Hai Precision Industry, Tsing Hua University, Brother Kogyo Kabushiki Kaisha, and others have caught up in some topics recently.
The results of the proposed methodology can assist policy makers in understanding technology trends and the innovativeness of assignees. This can aid them in identifying current trends and arranging development and deployment strategically in appropriate subdomains for policy making. In this study, characteristics were analyzed and concentrated to facilitate observing significant mean differences between transitions. The assignee level in the solar cell field was used as an example. The profiling results found that the endogeneity index and pendency period of the appearing assignees increased while those of the exiting assignees decreased, the originality index and science linkage of the appearing assignees increased while those of the exiting assignees decreased more than those of the stable ones, and the TCT of the exiting assignees increased while those of the appearing ones decreased. Such information potentially offers useful applications for policy makers to discern the status of an analyzed unit in an upcoming period. The proposed methodology can support analysis at different levels (e.g., inventor, assignee, and country) for various topics. It also enables forecasting an analyzed unit’s transition according to historical characteristic vectors by using a selected prediction tool. For each analyzed unit, the corresponding change rates in the most recent characteristic vectors could serve as signals for determining the analyzed unit’s transition to the next period. We suggest that future research elucidate more characteristics related to transition patterns to explore intelligence from different perspectives.
This research was partially supported by the National Science Council, Taiwan, under contract no. NSC102-3113-P-002-029. The authors also express their gratitude to Dr Ssu-Han Chen and Mr Ming-Hui Wu for their support in collecting data.
Aharonson B. S. & Schilling M. A. (2016). Mapping the technological landscape: Measuring technology distance technological footprints and technology evolution. Research Policy45(1) 81-96.
Ali A. Krapfel R. & LaBahn D. (1995). Product innovativeness and entry strategy: impact on cycle time and break-even time. Journal of product innovation management12(1) 54-69.
Boyack K. W. & Klavans R. (2008). Measuring science–technology interaction using rare inventor–author names. Journal of Informetrics2(3) 173-182.
Breitzman A. F. & Mogee M. E. (2002). The many applications of patent analysis. Journal of Information Science28(3) 187-205.
Bruck P. Réthy I. Szente J. Tobochnik J. & Érdi P. (2016). Recognition of emerging technology trends: class-selective study of citations in the US Patent Citation Network. Scientometrics107(3) 1465-1475.
Caillaud F. & Ménière Y. (2014). Strategic intelligence on patents. In Madiès T. Guellec D. & Prager J. C. (Eds.) Patent Markets in the Global Knowledge Economy: TheoryEmpirics and Public Policy Implications (pp. 33-54). Cambridge University Press.
- Export Citation
Caillaud, F. & Ménière, Y. (2014). Strategic intelligence on patents. In Madiès, T., Guellec, D., & Prager, J. C. (Eds.),)| false Patent Markets in the Global Knowledge Economy: Theory, Empirics and Public Policy Implications(pp. 33-54). Cambridge University Press. 10.1017/CBO9781107110960.003
Chen D. Z. Chang H. W. Huang M. H. & Fu F. C. (2005). Core technologies and key industries in Taiwan from 1978 to 2002: A perspective from patent analysis. Scientometrics64(1) 31-53.
Chen S. H. Huang M. H. & Chen D. Z. (2012). Identifying and visualizing technology evolution: A case study of smart grid technology. Technological Forecasting and Social Change79(6) 1099-1110.
Chen S. H. Huang M. H. Chen D. Z. & Lin S. Z. (2012). Detecting the temporal gaps of technology fronts: A case study of smart grid field. Technological Forecasting and Social Change79(9) 1705-1719.
Ernst H. (1999). Key inventors: Implications for human resource management in R&D. In Management of Engineering and Technology 1999. Technology and Innovation Management. PICMET’99. Portland International Conference on (pp. 420-427). IEEE.
Fan X. Liu W. & Zhu G. (2017). Scientific linkage and technological innovation capabilities: international comparisons of patenting in the solar energy industry. Scientometrics111(1) 117-138.
Haupt R. Kloyer M. & Lange M. (2007). Patent indicators for the technology life cycle development. Research Policy36(3) 387-398.
Hirschey M. & Richardson V. J. (2001). Valuation effects of patent quality: A comparison for Japanese and U.S. firms. Pacific-Basin Finance Journal9 65–82.
Huang M. H. & Chang C. P. (2015). A comparative study on detecting research fronts in the organic light-emitting diode (OLED) field using bibliographic coupling and co-citation. Scientometrics102(3) 2041-2057.
Huang M. H. Chen S. H. Lin C. Y. & Chen D. Z. (2014). Exploring temporal relationships between scientific and technical fronts: A case of biotechnology field. Scientometrics98(2) 1085-1100.
Huang Y. H. (2009). An investigation of the interaction between science and technology from literatures of giant magnetoresistance. Journal of Library and Information Studies7(1/2) 53-76.
Kayal A. A. (1999). Measuring the pace of technological progress: Implications for technological forecasting. Technological Forecasting and Social Change60(3) 237-245.
Lee S. Yoon B. Lee C. & Park J. (2009). Business planning based on technological capabilities: Patent analysis for technologydriven roadmapping. Technological Forecasting and Social Change76(6) 769-786.
Leydesdorff L & Rafols I. (2011). Indicators of the interdisciplinarity of journals: Diversity centrality and citations. Journal of Informetrics5(1) 87-100.
Liu J. S. Kuan C. H. Cha S. C. Chuang W. L. Gau G. J. & Jeng J. Y. (2011). Photovoltaic technology development: A perspective from patent growth analysis. Solar Energy Materials and Solar Cells95(11) 3130-3136.
Lo S. S. (2010). A comparative study of linkage indexes: Co-assignee reciprocal citation patent coupling and co-patent. Journal of Library and Information Studies8(1) 11-27.
Madani F. & Weber C. (2016). The evolution of patent mining: Applying bibliometrics analysis and keyword network analysis. World Patent Information46 32-48.
Narin F. (1991). Globalization of research scholarly information and patents–ten year trends. The Serials Librarian21(2-3) 33-44.
Narin F. (2000). Tech-line® background paper. In Tidd J. (Ed.) From knowledge management to strategic competence (pp. 155–195). London: Imperial College Press.
Newman M. E. J. (2004). Analysis of weighted networks. Physical Review E 70(5) 056131.
Park G. Shin J. & Park Y. (2006). Measurement of depreciation rate of technological knowledge: Technology cycle time approach. JSIR Vol.65(02). Retrieved from http://nopr.niscair.res.in/handle/123456789/4806
Pilkington A. Lee L. L. Chan C. K. & Ramakrishna S. (2009). Defining key inventors: A comparison of fuel cell and nanotechnology industries. Technological Forecasting and Social Change 76(1) 118-127.
Trajtenberg M. Jaffe A. & Henderson R. (1997). University versus corporate patents: A window on the basicness of invention. Economics of Innovation and new technology5(2) 19-50.
Upham S. P. & Small H. (2010). Emerging research fronts in science and technology: Patterns of new knowledge development. Scientometrics83(1) 1588-2861.
USPTO (2009). Pilot program for green technologies including greenhouse gas reductions. 74 Fed. Reg. 64666.
Zhao D. & Strotmann A. (2014). The knowledge base and research front of information science 2006–2010: An author cocitation and bibliographic coupling analysis. Journal of the Association for Information Science and Technology65(5) 995-1006.