A Predictive Model of Technology Transfer Using Patent Analysis

A Predictive Model of Technology Transfer Using Patent Analysis

The rapid pace of technological advances creates many difficulties for R&D practitioners in analyzing emerging technologies. Patent information analysis is an effective tool in this situation. Conventional patent information analysis has focused on the extraction of vacant, promising, or core technologies and the monitoring of technological trends. From a technology management perspective, the ultimate purpose of R&D is technology commercialization. The core of technology commercialization is the technology transfer phase. Although a great number of patents are filed, publicized, and registered every year, many commercially relevant patents are filtered through registration processes that examine novelty, creativity, and industrial applicability. Despite the efforts of these selection processes, the number of patents being transferred is low when compared with total annual patent registrations. To deal with this problem, this study proposes a predictive model for technology transfer using patent analysis. In the predictive model, patent analysis is conducted to reveal the quantitative relations between technology transfer and a range of variables included in the patent data.

Introduction

It has become important to promote open innovation, since growing uncertainties in the world economy have seen a contraction of the technology market [1,2]. The R&D environment has changed accordingly; it thus requires new methods of technology analysis and related analytical tools. It is important to understand that the new R&D environment is concerned not only with marketing strategies based entirely on product development and sales, but also with the marketing of intangible R&D-based assets such as intellectual property. As patents have become important tools in creating economic profits, it is necessary to invest resources in establishing patent strategies [1]. Many countries, such as the United States, the European Union, Japan, China, and Korea, have programs to protect intellectual property and to promote technology transfer and commercialization. Although the number of patents applied for and registered is rapidly increasing every year [3], there is still a disproportionately low number of technology transfers. Many countries are attempting to activate technology transfers by supporting patent offices and providing credit guarantee funds, but these efforts rarely contribute towards increasing the number of technology transfers. This is because there are several problems in selecting core patents to commercialize. There are few objective indicators or systematic approaches for identifying high-quality patents. Expert opinions are usually relied upon to select patents for technology transfer. This can only be effective in expanding technology transfers if the relevant expert has deep knowledge of the related technological areas. However, it is impossible to match appropriate experts with the huge number of patents filed every year. Because of the vast amount of patent applications, it is possible that a patent representing core technology can be poorly evaluated by experts with no background knowledge of the specific domain. Conversely, a patent with no technological potential could be estimated as a high-quality candidate for technology transfer and commercialization. These faulty evaluations result in a waste of R&D resources such as time, effort, and cost. In the process of extracting high-quality patents, experts generally get information from patents including basic technical information such as applicant, inventor, application date, abstract, statement, claim, figures, and so on. This research proposes a patent-based predictive model of technology transfer, obtaining quantitative relationships between major influential factors and successful technology transfers. This proposed predictive model is useful not only for predicting technology transfers, but also for preventing mismatch errors from expert opinions and the waste of R&D resources. The predictive model is constructed by preprocessing patent data and performing social network analysis, linear regression analysis, and decision tree modeling.

Related Work

Patent information analysis enables the extraction of a high-quality patent by analyzing its novelty, marketability, life cycle, citation information, and so on. As previous studies show, technology transfer has normally been researched through survey-based statistical analyses of subjective decision-making processes so as to propose qualitative strategies of technology transfer. Studies on patent indicators have mainly been concerned with forecasting promising technologies and patent registrations. Our approach, which applies quantitative analysis for technology transfer prediction, has not previously been investigated in the literature.

Predictive Model

With intensified global competition and the increased pace of technological development, it is vital to maintain competitiveness through effective innovation. Companies develop technologies through in-house R&D, but new technologies may also be acquired via open innovation technology transfers. Our aim is to support efficient R&D management through the use of predictive models based on statistical analysis and machine learning. This predictive approach should promote business sustainability through active technology transfers. Sohn and Moon [4] suggested a predictive structural equation model that creates a technology commercialization success index (TCSI). Although many companies invest in R&D, they suffer asset losses when these efforts cannot be commercialized. To reduce these risks, this research applied a structural equation model for predicting TCSI values, given the technology receiver, technology transfer center, and environmental factors. As a follow-up study, Sohn and Moon [5] proposed a decision tree model based on data envelopment analysis (DEA) to evaluate effective technology commercialization. In the case of IT companies attempting technology transfer or commercialization, DEA findings suggested the decision tree model as an appropriate tool in developing effective project roadmaps. Hwang and Lim [6] conducted research on selecting the best R&D scenario by combining Monte-Carlo simulation and decision tree modeling. Walker et al. [7] undertook a comparative analysis of various types of future planning approaches to deal with technological uncertainty and suggested several methodologies to implement these approaches. Many previous studies have applied technology indicators or technology transfer algorithms to enable stable technology commercialization and transfer. There has been little research on predictive models of technology transfer using patent analysis. That is, traditionally the domain experts have decided to the possibility of whether the transfer of technology. But this is subjective and not stable. To solve this problem, we use objective method using quantitative approach based on patent analysis and statistical methods.

Patents and Patent Analysis

Patent systems function to grant inventors the exclusive right to own their inventions [8]. Inventors specify the technical details and experimental results of their inventions in order to protect their exclusive rights. Patent documents are thus a valuable source of information on a wide range of developing technologies. Choi et al. [9] conducted a co-classification analysis of all patent data submitted to the Korea Intellectual Property Office from 1988 to 2010. By analyzing trends of technological convergence, it was found that the number of patents diminishes as convergence grows rapidly. Yoon and Park [10] suggested a hybrid approach to patent analysis by combining conjoint analysis and patent citation information. This approach was applied to the thin film transistor liquid crystal display patent database as an empirical study. Jun and Park [11] conducted quantitative patent analysis using statistical methods, machine learning, and social network analysis to analyze patent data from Apple. The study proposed a methodology for extracting Apple’s technology trends and relationships among various technologies; it resulted in extracting vacant technology areas of Apple. Mogee [12] used patent families to analyze R&D planning, international patent activity, and patent indicators. The resulting growth model revealed distinct differences between mature and promising technologies. Wu et al. [13] constructed a predictive model based on international patent classification (IPC) codes to assess possibilities of patent registration. A genetic-based support vector machine was used to construct a model to screen patent data as an alternative to relying on expert opinions. Park et al. [14] constructed an expert system to predict patent registration, using a model based on quantitative methods such as patent indicators and text-mining techniques. Jun et al. [15,16] studied data mining techniques for technology forecasting using patent information such as title, abstract, IPC codes, and bibliographic data.

Technology Transfer

Anderson et al. [17] evaluated technology transfer production from U.S. universities using the DEA approach. DEA was specifically applied to evaluate technology transfer offices (TTOs), using the number of technology transfers and their effectiveness. The study also employed linear regression to find out whether medical school is an essential factor for a TTO’s efficiency. Weckowska [18] conducted an empirical study of TTOs at six universities in the United Kingdom to analyze two different types of technology transfer. Chang and Chen [19] discussed potential applications from fuzzy set theory to select biotechnology management strategies and technology transfers. Their algorithm was based on the concepts of fuzzy set theory and hierarchical structure analysis. Linguistic variables and fuzzy number were used as weighted evaluation values along with the subjective evaluations of decision makers. Heslop et al. [20] designed a preparatory stage for technology transfer, known as the “Cloverleaf Model,” to achieve successful transfers. Cummings and Teng [21] demonstrated that successful technology transfer is related to various factors such as a company’s understanding of R&D, its expertise in technology transfer, and the range of shared basic knowledge. Mowery et al. [22] examined technology transfer within an organization for strategic linkages. The study extracted a new indicator that reflects changes in technological ability based on the citation pattern of a patent portfolio. In addition, there are many existing studies such as technology transfer monitoring for government-funded R&D projects [23], technology transfer as a learning and developmental process for Norwegian programs [24], and analytic hierarchy process modeling to evaluate the indicators of technology transfer strategies in the petrochemical industry [25]. Kim [26] proposed a hierarchical analysis model of the decision making process in order to predict technology transfer policy directions. Sohn et al. [27] analyzed the technology market’s environmental factors to narrow the gap between demand and supply and to activate technology commercialization.

Predictive Model of Technology Transfer

In general, the transferred technologies are important to a company and nation for improving their technological competitiveness. In addition, it is needed that the technologies are researched and developed sustainedly. There is close correlation between technology transfer and sustainable technology. Thus, we can forecast the sustainability of a technology by predicting the possibility of technology transfer. This research proposes a predictive model of technology transfer by collecting patent data and applying text mining techniques for preprocessing. The model then performs social network analysis, regression analysis, and decision tree modeling.

Further read…

Department of Industrial Management Engineering, Korea University, Seoul 136-701, Korea

Department of Statistics, Cheongju University, Chungbuk 363-764, Korea

Graduate School of Management of Technology, Korea University, Seoul 136-701, Korea