Technology Early Warning Model: a New Approach Based on Patent Data Ganlu Sun, Ying Guo Fan Yang Beijing Institute of Technology WIDE-CODE Information Technology Co.,Ltd 5 South Zhongguancun Street Wangjing High-tech Park, Lize Zhong Er Road, Haidian District Chaoyang District Beijing, P.R.China Beijing, P.R.China +8613810027376 (86)(10)64878770 sunganluzz@163.com, guoying_bit@163.com paoever@163.com ABSTRACT (Congmin She.2003) [2]. This can offer enough time for early With the development of technology, more and more technical warning subjects to take mearsures before crises happen. It also issues have been exposed, such as technical disputes, technical can help early warning subjects to reduce significant loss as a barriers and technical crisis. Thus, it is necessary to warn result. enterprises about technical deviation and predict future The concept of technology early warning was first put forward by technology crises. Patent data can contain much information the U.S. military. In this concept, technology early warning refers about technologies and would be useful in this setting. This to the alertness to a “technical raid” which potential rivals may paper proposes a technology early warning model based on form in advance to keep military advantage in technology (Boao patent data. This model helps enterprises analyse the technical Qin.2006) [3]. Yang Cai (1989) [4] redefined it, pointing out that crisis level and trends from four different perspectives (technical technology early warning is a process from technology forecast stability, technical monopoly, technical security and technical and relevant factors’ breakthrough to give the alarm to decision prospects). makers. Yujie Zhang (1999) [5] defined technology early warning in enterprise as a security alarm of technical deviation and Keywords technical catch up situations which remind enterprises take Technology early warning, Patent data, Technical crisis, measures to keep a technical advantage. In this paper, technology Indicators early warning is regarded as forecast and alert for technical crisis which would threaten enterprises’ sustainable development, and lose their technological advantage. 1. INTRODUCTION With the development of science and technology, many In order to execute technology early warning, technology enterprises, especially high and new technology enterprises, information needs to be considered. Many enterprises apply for started to focus on research and development of technology. patents to protect their technology infringment by other actors, However, as a result of the development of economic and are effective means to protect intellectual property. Recent globalization, there are many technology disputes between studies have used patent information to study technological enterprises all over the world every year. Thus, it is necessary for developments, trends and potential (Zhang, 2011; Pilkington et enterprises to take some measures to avoid these dispute and al., 2009) [6-7] as well as decision making in research and keep their technology advantage in a fiercely competitive market. development (Thorleuchter et al., 2010) [8]. Information from We address this in our paper and we aim to provide an effective patent data allows companies to avoid investing in obsolete method for enterprises in technology early warning. technology (Wang et al., 2012) [9] and it enhances strategic planning (Abraham and Moitra, 2001) [10]. Patent analysis also The phrase early warning derives from military planning, offers key information concerning the technology environment refering to predicting enemy attacks, giving the alarm in a timely (Porter and Cunningham, 2005) [11] and addresses component manner and preparing the appropriate response to avoid technologies (Trappey et al., 2012) [12]. Patent data is suitable to significant loss (Jiezhu Pan, 2007) [1]. In other fields, early analyze technology in enterprises and this paper will develop warning is a forecast method for crisies which could threaten that patent analysis further by introducing perspectives on technology field’s normal operation, and offer directions for preparation early warning. Currently, there are few studies which incorporate technology Copyright © 2015 for the individual papers by the papers' authors. early warning with respect to technical crisis and combine this Copying permitted for private and academic purposes. with patent data. Therefore, we propose a new method for the This volume is published and copyrighted by its editors. technology early warning model based on patent data considering Published at Ceur-ws.org Proceedings of the Second International Workshop on Patent Mining and forecasting of technical crisis. We aim to contribute to improve its Applications (IPAMIN). May 27–28, 2015, Beijing, China. technology early warning systems. This paper identifies, analyzes, and indicates the technology crisis, and then establishes the technology early warning indicators and a model based on patent data. We aim to provide a new method for enterprises to which are “health care”, “computer technology”, “management” solve problems in technical disputes, technical deviation and and “perception”. However, as we can see in Figure 1, “GPRS” technical catch up situations before they happen. and “early warning score” are the main research domains within This paper is organized as follows. The literatures about management science. However, this paper will address different technology early warning are reviewed in Section 2. Model of perspectives in “technology early warning” and combine this technology early warning based on patent data are proposed in with technical crisis values. To learn more about research in Section 3. Conclusions and future research directions are management science, we ranked papers’ keywords which appear provided in Section 4. more than twice – see Table 1. From Table 1 we observe that much research focuses on early warning methods or systems. 2. LITERATURE REVIEW These studies utilize computer technology (such as data mining To establish a better technology early warning model, we and AHP) to simulate or forecast alert situations. From these experimented with information on technology early warning keywords, we also build on the quantifiable methods used by research status from a broad perspective in a bibliometric many of these papers in our research. database. This gave us a perspective on the theoretical concepts In tandem with publication data considering technology early and current trends in the literature of early warning. Our results warning models, we addressed the development of technology show a wide range of information about the current research early warning in a patent database. Initially, we searched in fields, previous methods and current critical dialogues of “Web of Science” with a boolean search term- “Topic= (patent) technology early warning research. AND Topic= (technology early-warning OR technology early Firstly, we searched in “Web of Science” using a boolean search warn OR tech early-warning OR technology early warning OR term-“Topic= (technology early-warning OR technology early technical warning)” and generated 8 results, with none related to warn OR tech early-warning OR technology early warning OR management science. We altered the ‘topic’ term to “patent AND technical warning)” to study the current research trends and early warning” and generated 17 results, with 5 papers explicitly indicate the leading field of technology early warning researchs. mentioning patent data in early warning but no reference was We generated 3436 results which indicated a sufficent quantity made to technology early warning. Of the total, four papers of research to be worth analysing in a bibliometric database. where relevant for our study.One study, commented on the After that we downloaded these documents and processed them research literature of early warning mechanisms in China, and using VantagePoint (a software for data processing), locating 135 presented an international case study of a successful early items related to management sciences. We can observe that while warning technique using patents (Jianping G, 2011)[13]. Another there have been many studies ontechnology early warning, there paper discussed the importance of establishing patent early have been fewer in management sciences and is worth warning systems to forecast potential patent risks and analyzed considering in the context of enterprise readiness for technical the main reason of patent risk. It proposed three basic functions crisis. . of a patent risk early warning system. This paper also presented a basic framework and model of that system (Han, Hongqi; Wang, We extracted keywords in 3436 papers and selected the top 200 Xuefeng, 2008) [14]. Another paper proposed patent infringement which were clustered in a visualisation- see Figure 1. From litigation early warning indicators and a model for Figure 1 we can see that keywords in previous technology early pharmaceutical enterprises. warning research has mainly clustered into four categories Table 1. Keywords in management rank Times keywords 18 Early warning system 17 Early warning 10 Technology 7 Technology Innovation 6 Risk management 5 Classification; Crisis; Data mining; Discriminant-analysis; Management; Model 4 China energy; Complex adaptive system; Construction project; Consumption; Early warning management; Emergency; Indicator system; Information; Real estate; systems; The ability of independent innovation; Trends; Warning system 3 Admissions; AHP; Fuzzy comprehensive evaluation; Governance; Information extraction; Intensive-care; Intrusion detection; Lessons; Logit; Network; Antecedents; Competition; Data and information quality; Deaths; Design science; Early warning score; Environment early-warning; Failure; Financial crisis; Neural networks; Project management; Risk assessment; Sensor-based electronic modified early warning scorecard; Socio-technical information systems design methodology; Stakeholders; Strategy; The model of early-warning Factor Map Keywords (author's) + Keyword... public health Kenya Floodvegetation Factors: 30 % Coverage: 33% (1141) forecasting Top links shown > 0.75 0 (0) Ethiopia 0.50 - 0.75 0 (0) 0.25 - 0.50 0 (0) EL-NINO < 0.25 27 (201) malaria localization uncertainty CARE PCR pollution SCIENCE hazard Health Technology Assessment PCR vulnerability DISCRIMINANT-ANALYSIS Neural network Data fusion IMPACT early warning score environment INFECTION management GPRS perception Magnitude Copper Wireless sensor surgery networks Figure 1. Top 200 keywords’ cluster This also presented a three stage early warning indicator system and gradualness (Lixin Xia et al., 2009[17]; Runhua Tan, 2002 (Wang, Ying, 2013) [15]. The final paper considers how the patent [18] ). To forecast technical crisis, we should consider the early warning mechanism can improve innovation abilities of characteristics. (1)Uncertainty of technical crisisimplies the risks wind power enterprises. They imply that by establishing patent involved in a period of development. Characteristics of the early warning mechanisms, common issues experienced by technology, internal and external influences, and limitation of enterprises in the sector can be resolved through these methods human knowledge could increase uncertainty in these terms. (Peng Yuanyuan, 2012)[16]. These articles imply that while there (2)Relativity of technical crisis is caused by different is growing dialogue on the subject, research on technology early organizational conditions. The same technology may have warning combined with patent data is an underdeveloped area diffierent risks in different organizations as well as in the same worthy of further study. organization, at a different stage of development. Thus, technology crisis may be both an opportunity and a challenge for 3. METHODOLOGY an enterprise. (3)Destructiveness of technical crisis implies that 3.1 Research Framework once a crisis begins, it can cause serious disruption for an In this paper, we regard technology early warning as an alert for enterprise. (4)Imperceptibility of technical crisis implies an event technical crisis. Technical crisis is a risk that caused by the which cannot be detected and may have significant affect on the accumulation of negative factors of technologies’, and it will enterprise. This will contribute to the destructive nature of destroy the quality of a technology if unaddressed. Technology technical crisis. (5)Gradualness of technical crisis implies the has features of stability, monopoly, security, reliability and accumulative quality of negative factors in the technology and development potential. However, when technologies are technical environment. This conceptual outline is demonstrated destroyed, it will turn into a technical crisis for the enterprise in Figure 2. causing uncertainty, relativity, destructiveness, imperceptibility Technical Technical crisis characteristic characteristics s Stability Uncertainty Monopoly Relativity Destroy Technology Security early warning Destructiveness Reliability Imperceptibility Prospects Gradualness Figure 2.Triggers of technical crisis With a well defined characteristic model of technical crisis, we  How will enterprises’ technology develop in the future? can further develop our analysis towards a technology early warning model. Judging technical crisis through enterprises’ 3.2 Indicators data directly is a complicated process, so we reflect on thus by In order to answer these questions and help enterprise using information gained from the technologies. Therefore, this forecast its technical crisis with quantitive methods, we propose paper will execute technology early warning research through the four different perspectives (technical stability, technical patent data which displays information about enterprises monopoly, technical secuity and technical prospects) indicators to technology development. This paper aims to answer the analyze enterprise technology status. From this we will judge following research questions with patent data in enterprises: whether an enterprise will encounter technical crisis. From this we can present an ‘early warning’ index for companies. Based  How do we judge whether an enterprise technology will be on the rich literature review and consultation with experts, we affected by internal and external factors? defined ten indicators to measure these different perspectives.  At what level is an enterprise technology in its field? Indicators are shown in Table 2.  Can enterprises protect their technologies? Table 2. Indicators and its detailed contents Perspectives Indicators Operational definition Technology maturity Enterprise technology’s development stage Technical R&D staff flow Change of enterprise R&D staff stability The degree of enterprise technology dependent on Technical dependence external factors Research scope of enterprise technology (technical Technical breadth Technical breadth) monopoly Enterprise technology’s strength in research Technical strength (technical depth) Technical disparity Differences of enterprise technology Technical security Technical complexity Complexity of enterprise technology system Technical maintenance Ability to maintain technology Technical Technical progress Development trend of enterprise technology prospects Technical environment Development trend of the technical field First, the perspective of technical stability is used to judge if enterprise technology can develop stability and to what extent it Table 3. Computational formula of each indicator in first will be affected by internal and external factors. This insight is perspective gained from technology maturity, technical staff flow and technical dependence. Technology maturity reflects technical Indicators Formulas Explaination stability mainly through technology life cycle. R&D staff flow We define the value of reflects technical stability through the number of R&D workers We calculate Technology maturity stage which (can be called ‘technicists’) who apply for a patent in an maturity based on a is the most stable enterprise. We consider the more technicists are engaged in these Technology growing model and stage as 5, and the activites, there is a higher stability coefficient in the enterprise. maturity divided this into five stage closer to Technical dependence is a powerful indicator to reflect technical levels. maturity stage have stability. We argue that if enterprises have a significant higher value. dependence on another enterprise technology, they are more prone to technical crisis. The computational formula of each AI: average inventors indicator is shown in Table 3. in patent applications. R&D staff TI TI: total inventors to The second perspective--technical monopoly--is used to judge at AI   100% flow TPE apply for patents. what level an enterprise technology is in its field. To achieve TPE: total patents this, we consider technical breadth and technical strength to enterprise has. measure technical monopoly. We introduce this indicator to measure research scope of enterprise compared with its rivals TDE: technology and we measure technical breadth as the proportion of the dependence. TPF: enterprise’s technology accounted for in its field. Technical TPF  TPE total effective Technical TDE   100% strength reflects technical monopoly in the view of an TPF invention patents in dependence enterprise’s strength in research and development of a the field. TPE has the technology. We argue that the value of technical strength has a same meaning with positive influence on technical monopoly. In Table 4 we detail above. the computational formula of these indicators. Table 4. Computational formula of each indicator in second Finally, the perspective of technical prospects is proposed to perspective estimate the development prospects of an enterprise technology. We introduce technical progress and technical environment to Indicators Formulas Explaination measure this. Technical progress reflects technical prospects IPC q TB: technical breadth. IPCq: from the variation of patent numbers. We argue that technologies Technical TB   100% number of IPC in enterprise. will develop well in the future when patent numbers increase breadth IPC h IPCh: the number of IPC in rapidly. The technical environment reflects technical prospects of enterprise’s industry. the field and how the enterprise is affected by it. We argue that TS: technical strength. NAIP: an enterprise technology will have the same potential of the number of active invention development as the field it inhabits. Computational formula of Technical TS  NAIP  ALPE each indicator is shown in Table 6. patent in an enterprise. ALPE: strength average length of patent enforcement. Table 6. Computational formula of each indicator in third perspective The third perspective--technical security--is used to measure an Indicators Formulas Explaination enterprise’s capacity to protect its technology from damage. IAR: increase of Technical diversity, technical complexity and technical application rate for maintenance are indicators we use to measure technical securiy. invention patents. Technical diversity refers technology difference between NPC: number of enterprise and the field. And we argue that with a lower NPC  NPP patents by technical diversity value an enterprises technology is more Technical IAR   100% NPP enterprises in secure. Also, a higher value in technical complexity and technical progress current period. maintenance means a more robust enterprise technology. Table 5 NPP: number of shows detailed computational formula of these indicators. patents of enterprises in prior Table 5. Computational formula of each indicator in third period. perspective IRLP: the increase Indicators Formulas Explaination rate of active patents. NLPC: TDI: Technological number of active diversity. ai and bi NAPC — NAPP Technical IRAP   100patents of industry n means the number of environment NAPP in current year. Technical disparity TDI   b - a  i 1 i i 2 active invention patents in an NLPP: number of active patents of enterprise and its field industry in prior distribution in IPC year. categories, i=1,···,n. TC: technical complexity. NIPCC: 3.3 Process NAP the number of main Technical TC   100% NIPCC IPC categories. Based on indicators shown on 3.2, this method will also address complexity NAP:number of active the wieghting and degree of alertness which helps in the analysis patents in an of technical crisis. enterprise (1) Weighting AEG: average AHP (Analytic hierarchical Process) is a powerful decision TNIP effective age of analysis technique for multi-criteria decision-making, and it can Technical AEG   100% technology. TNIP: the decompose problems into a hierarchy of goals, attributes, criteria main- NAP total number of and alternatives. Therefore, after the calculation of each indicator tenance invention patents in an with patent data in 3.2, we used AHP to weight each indicator. enterprise. Table 7 is the result of the weighting of each indicator. Table 7. Summary list of the indicators Technical Technical Technical Technical Synthetic Rule level stability monopoly security propsects weight Indicators 0.0507 0.1781 0.5539 0.2172 Technology maturityC11 0.6434 0.0326 R&D staff flowC12 0.0738 0.0037 Technology dependence 0.2828 0.0143 C13 Technical 0.1667 0.0297 widthC21 Technical strengthC22 0.8333 0.1484 Technical diversityC31 0.1530 0.0847 Technical complexityC32 1.0548 0.5843 Technical maintenanceC33 0.4010 0.2221 Technical progressC41 0.8333 0.1810 Technical environmentC42 0.1667 0.0362 (2) Setting alert degree In this model, we calculate the total evaluation score by multi- Wherein T means the degree of enterprise technical crisis, m objective linear weighting function. Formula of enterprise means the number of perspectives, n means the number of technical crisis is shown as follows. indicators in each perspective. And Wi means the weight of ith perspective.Wij is the weight of jth indicator in ith perspective, Rij is the grade of jth indicator in ith perspective. We outline m n degree of alertness in Table 8. Enterprises can know their T   R W i  1 ~ m; j  1 ~ n i 1 Wi j 1 ij ij technical crisis level and safety situation by the value of T contrasted with Table 8. 4. CONCLUSIONS The objective of this study is to propose a new approach of Table 8. Alert Degree and its meaning technology early warning with patent data. To this end, we Degree of analyzed existing literature in the Web of Science to find the Score Meaning research situation of technology early warning and some studies technical crisis combined with patent data. Based on the previous studies of Safest Technical crisis in enterprise is technology early warning, we proposed four different 0~1 slight, and it will not affect （A） perspectives to consider patent and technical crisis characters enterprise’s interests and actions. and defined ten indicators to analyze these perspectives. After Safer Technical crisis is not serious, and that we analysed the weighting of each indicator, and set the 1~2 it will only cause minor damage on degree of alertness to measure the degree of technical crisis （B） the enterprise. theenterprise faced. Enterprise technology has a certain This study offers some direction for forcasting enterprise Safe crisis, and it will affect the technical crisis using patent data. The model for technology early 2~3 （C） enterprise’s interests and actions warning can be used to reduce technical disputes and technical but will not be fatal. barriers as well as aid further academic research. These models could be adopted by Enterprises with relevant patent data, to Technical crisis in an enterprise will Risky analyze technical crisis and better react to it. bring cause a signifcant loss. The 3~4 （D） possibility of technical disputes will As with any experimental model there are limitations which will rise. need to be improved. The most important one is that as the reason of time and difficult in getting patent data for us, we were Highly Risky Technical crisis is inevitable, and it unable to verify the index system and model by empirical study. 4~5 may directly threaten the （E） So in the future research, we will do an empirical study to prove enterprise’s survival. the feasibility and rationality of this new method and make it better. Because of much knowledge involved in this research, identifying relationships between technologies. Technol. there are some difficulties caused by abstraction of research Forecast. Soc. Chang. 77 (7), 1037–1050. contents that make our indicators fuzzy. In the future, another [9] Wang, Y.L., Huang, S., Wu, Y.C.J. 2012. Information work for us is to research the method of indicator quantitative. technology innovation in India: the top 100 IT firms. Technol. Forecast. Soc. Chang. 79 (4), 700–708. 5. ACKNOWLEDGMENTS [10] Abraham, B.P., Moitra, S.D. 2001. Innovation assessment This work was supported by MOE (Ministry of Education in through patent analysis. Technovation 21 (4), 245–252. China) Project of Humanities and Social Sciences (Award [11] Porter, A.L., Cunningham, S.W. 2005. Tech Mining: No.13YJC630042) and Key Projects in the National Science & Exploiting New Technologies for Competitive Advantage. Technology Pillar Program (2013BAH21B00). John Wiley & Sons, Inc., Hoboken, New Jersey. [12] Trappey, A.J.C., Trappey, C.V., Wu, C.Y., Lin, C.W. 2012. A patent quality analysis for innovative technology and 6. REFERENCES product development. Adv. Eng. Inform. 26 (1), 26–34. [1] Jiezhu Pan. 2007. Research on Technology of Early Warning Based on Data Mining. Hefei University of [13] Jianping G.2011. Research on Constructing Effective Patent Technology, P.R.China:2-10. Early Warning Mechanism in Central Region of China. Yamaguchi Univ, Kitakyushu, JAPAN: proceedings of the [2] Congmin She.2003.Forecasting, P.R.China: Theory 8th international conference on innovation and management, Summarize of Research on Enterprise Early Warning in 949-955. China.22 (2):23-30. [14] Han, Hongqi; Wang, Xuefeng.2008. Study on the model [3] Boao Qin.2006. Experts in China Defense Science and construction of early warning system of patent risk. Beijing Technology Information Center analyze “technology rid” Inst Technol, Sch Management & Econ, P.R.China: and “technology early warning”. P.R.China: Modern proceedings of 2008 international conference on risk and Military.2, 12-16. reliability management, 408-411. [4] Yang Cai, Qingming Huang.1989. Analyze of the [15] Wang, Ying.2013. Construction of Patent Infringement characteristic of technology early warning. P.R.China: Litigation Early Warning Index System for the Forum on Science and Technology in China. 5,58-59. Pharmaceutical Enterprises. P.R.China: Materials, [5] Yujie Zhang.1999. Technology early warning of enterprise. mechanical engineering and manufacture, 268-270. P.R.China: Journal of Beijing Institute of Planning Labour [16] Peng Yuanyuan.2012. Discussion about Early Patent Administration General. 7(3), 41-42. Warming Mechanism of Chinese Wind Power Enterprises. [6] Zhang, L. 2011. Identifying key technologies in P.R.China: Advances in computer science and engineering, Saskatchewan, Canada: evidence from patent 769-775. information.World Patent Inf. 33 (4), 364–370. [17] Lixin Xia, Jun Wang.2009. The research on enterprise [7] Pilkington, A., Lee, L.L., Chan, C.K., Ramakrishna, S. technical crisis early warning based on competitive 2009. Defining key inventors: a comparison of fuel cell and intelligence. P.R.China: Practice research.32 (9), 68-71. nanotechnology industries. Technol. Forecast. Soc. Chang. [18] Runhua Tan.2002. Innovative design---TRIZ: the theory of 76 (1), 118–127. solution in problems of invent. China Machine Press, [8] Thorleuchter, D., Poel, D.V., Prinzie, A. 2010. A compared Beijing. R&D-based and patent-based cross impact analysis for