ABSTRACT. Scientific breakthroughs play a key role in the development of the field. Exploring the development laws of such topics will help to carry out subsequent identification and prediction, and can make scientific layout earlier and optimize scientific and technological decisions. This article took the field of genetic engineering vaccines (GEV) as an example to carry out empirical research. Starting from the basic point that scientific breakthrough topics will affect the state of the knowledge network, the topic words co-occurrence network is used to represent the knowledge network, and the structure entropy is used to measure the state of the knowledge network. Afterwards, discover topics that have a greater impact on the state of the knowledge network, as candidates for scientific breakthrough topics, and identify the effects with the help of experts and existing scientific evaluation methods. The structural entropy index is used to measure the overall change of the knowledge network, which is suitable for the identification of scientific breakthrough topics, but the stability of its effect and whether it can be used in other fields needs further verification and exploration.

ABSTRACT. Anticipating promising patents is of essential importance for companies, as it may reveal opportunities to enter a novel market or upcoming threats for incumbent firms. However not every patent is economically relevant and valuable. Some scholars even argue that a significant fraction of patents is “worthless”. On top, the increasing number of patent applications per year makes it even more difficult for companies to identify promising patents. This insight leads to the demand for a method that can forecast promising patents. In this study, we propose an approach to identify promising patents. For this purpose, we combine link prediction algorithm with semantic patent analysis approaches and predict word combinations (so-called bigrams). For testing purposes, we apply our analysis to case of camera technology. As a result, we identify stochastic gradient descent as a suitable classifier with both a balanced accuracy performance and positive predictive value of 78 % outperforming chance by a factor between two and three. In addition, we identify promising patents addressing diverse application fields. In terms of practical implications, we assume that analysts can use our approach as an early warning system to detect promising patents immediately after publication. However, there is still a need for further research. For instance, we plan to validate our method by a comparative study and improve the model performance by varying our language pre-processing.

ABSTRACT. Theories of innovation point to the critical role of knowledge spillovers in driving innovation and economic growth. Yet, our understanding of the impact of external knowledge on progress in individual technology areas remains limited. This paper develops a novel method of identifying spillover knowledge flows across technology domains and investigating the sources, pathways, and impact of individual technology spillovers on innovation. First, we describe a conceptual framework that defines technology spillover as a boundary-spanning knowledge transfer event that drives innovation in the target technology domain. To operationalize the framework, we combine insights from qualitative methods, patent citation analysis, and a machine learning-based method of patent text analytics that tracks integration of knowledge represented by patents into the target domain. As part of this method, we construct a technological space for the technology of interest by creating an LDA topic model of all patent full texts belonging to the corresponding technology domain. We construct the technological trajectory as a sequence of “field centroids” – arithmetic means of all vectors representing patents in the topic space granted in a single year. Next, we identify backward citations from patents in the target domain to patents external to the domain as potential spillover events. For all such cited patents, we calculate their “novelty” and “relevance” relative to the technological trajectory as, correspondingly, their distance to and projection on the field centroid trajectory, and track the evolution of these metrics over time. We identify cited patents as representing important spillovers if they become less novel and increasingly relevant to the technological trajectory over time, thus showing gradual integration of the relevant technical knowledge contained in the patent into the mainstream of the technology. By combining this information with insights from qualitative methods such as archival research and expert interviews, we create detail-rich “spillover histories” and spillover knowledge network maps that track the role of technology spillovers in important innovations in the target technology, as well as the sources, channels, and enabling factors of these spillovers. We illustrate this approach by applying it to the case of crystalline silicon solar photovoltaics and demonstrate how it can generate insights for public policies and R&D management practices that explicitly harness spillovers to accelerate clean energy innovation.

ABSTRACT. Since this research seeks to improve AP of estimating the demographic characteristics of learners in MOOC platforms, we are aiming to further research in this area by studying traditional (textual) features of the subject. We have investigated the state-of-the-art literature in AP, which consisted at the time of Deep Learning models (DL), and thus, we are considering these models for our work. In the earliest stage of our research, we have focused on predicting the employment status of learners based on the semantic representation of text. We compared the sequential with the popular parallel ensemble deep learning architecture for AP based on Convolutional Neural Networks and Recurrent Neural Networks. We obtained an average high accuracy of 96.3% for our best method. On a following stage, we focused on predicting the gender of learners based on syntactic knowledge from the text. We have compared different tree-structured Long Short Term Memory models and provide our novel version of Bi-directional composition function for existing architectures. In addition, we have evaluated 18 different combinations of word-level encoding and sentence-level encoding functions on this. Based on these results, our Bi-directional version model outperforms in all models and the highest accuracy result among our versions models is the one that is based on the combination of Forward Neural Network and the Stack-augmented Parser-Interpreter Neural Network (82.60% predication accuracy). We show that a prediction model can achieve high accuracy, and that pre-course questionnaires with a high cognitive overhead for author profiling could become redundant.

ABSTRACT. We explore the use of large-scale and longitudinal textual analysis of historical sources to trace technological dynamics. The study of how technology evolves and triggers changes in the society and economy has been central to research efforts in science policy and innovation studies. Three main research streams have contributed to increase our understanding of this phenomenon. Scholars in economics of innovation have examined technological dynamics in terms of trajectories relying to a great extent on highly standardised patent data. Similarly, emerging technologies literature has focused on developing methodological approaches to map emergence using highly standardised bibliometric data. Both literatures have, however, neglected social and system-level processes, which is at the core of science and technology studies literature. Research in this area has focused on socio-technical systems and processes involved in transitions, using qualitative case studies. This has been, however, subject to critiques on the grounds of selectivity and subjectivity. Diverging research efforts on studying technological dynamics have resulted in a divide between which aspects of technology are `quantified' and `qualified'. We explore how increasing access to textual sources and advancements in text mining techniques represent an opportunity to bridge these literatures. To do so, we examine both technical and social aspects of the case of mass production using three sources that span a period of about 150 years: Scientific American, New York Times, and Google Books. We also explore methodological implications of the choice of sources and keywords as proxies for technological trends and their ability to observe historical trends.

ABSTRACT. This paper aims to analyze topics within the international classification of goods and services (NICE classes) applied for the registration of trademarks at the EUIPO. This is accomplished by introducing a more fine-grained classification of trademarks as a "sub-section" of the rather rough NICE classes. To do this, we relate the descriptions of the trademarks that the applicants provide upon filing to the list of pre-defined keywords that are available from the WIPO to assist the applicant in describing his or her mark. In order to relate the keywords to the classifications, i.e. to assign trademarks to the classification, we use two algorithms including a Levenshtein-based matching and a Jaro-Winkler algorithm based matching. The Levenshtein-based approach already leads to a coverage of 75% of matched trademarks. With the help of the Jaro-Winkler matching algorithm (in combination with the Levenshtein distance) we could assign another 10%, leading to a coverage of 85% of all EUIPO trademarks matched to at least one classification key in 2018. Based on this matching we generate a hierarchical classification including 5 layers, the first layer including 234 classes up to the 5th layer which comprises 8,613 distinct classes.

ABSTRACT. The transition towards a more sustainable future has become pivotal for the 21st century. In this sense the shift to a circular economy presents one option to meet the “Sustainable Development Goals”. The circular economy seeks to minimise waste generation and material inputs through eco-design, recycling and reusing of products. Here, numerous trends are emerging. Identifying and assessing trends and their scientific basis is important for companies, researchers and policy-makers to support the transition towards a circular economy. Fundamental to this task is the detection of weak signals and other indicators. Bridging information from trend data and scientific publications is a promising opportunity to identify and analyse the scientific basis of trends and to assess their importance in supporting the decision-making process. We collect title and abstract of publications dealing with circular economy and use the TrendOne database to collect similar data on trends associated with this topic. Two approaches are applied and compared. First, we bridge the two data sources using the doc2vec approach and apply an appropriate clustering approach to identify topics within this data set. The second approach is to cluster the data sets separately by doc2vec and e.g. k-means. We bridge the clusters by measuring semantic similarities with doc2vev. For both approaches we analyse the development of the clusters to identify the clusters evolutionary path. Current results reveal that trend data and scientific publications complement each other and clustering approaches based on semantic similarities provide insights into the scientific basis of trends and their development.

ABSTRACT. It has demonstrated that highly novel research leads to the increase of scientific impact in term of citation counts. An ongoing interest in technological innovation is to find out whether a technology with highly topical novelty results in more technological influence. Here, the topical novelty is quantified with the Citation Influence Model (CIM). Then, a negative binomial regression model is constructed based on several influencing factors. In the end, using a corpus of 1537 patents published between 2001 and 2010, we discover a positive relation between topical novelty and technological impact.

ABSTRACT. Nowadays, the patent war among enterprises is becoming more and more fierce. In order to gain advantages in the competition, most enterprises will bypass the patent barrier of others through patent design around, so as to open up new technology market faster by using existing patent technology. In addition, enterprises should also think about how to use patent enhanced design to prevent competitors from patent design around on their own patents, protect independent intellectual property rights, and enhance the independent innovation ability of enterprises. In order to effectively carry out patent enhanced design, firstly, analyze the structure of the patent claims and decompose the technical features, extract the key components of the system, analyze the functions and interactions of the system components, and then combine the patent enhanced design strategy to form a new technical solution and evaluate the solution, and finally verify the feasibility of the method through specific applications.

ABSTRACT. In the new stage of global technological competition, it is important to focus on key technologies and achieve technological innovation. Patent is the fastest, most comprehensive, and most systematic information resource in the world that reflects the development of science and technology. Through the analysis of patent, it can provide the necessary guarantee for technological innovation management. Therefore, this article takes patents as the starting point, integrates patent cluster analysis, technology efficiency matrix, technology roadmap and technology evolution trend and other methods and theories, builds a model of technology innovation based on patent analysis and technology integration, and identifies current technology hotspots from a micro level , gaps and development trends. Finally, taking the technological evolution route of identifying the gas turbine casing as an example, it shows that this model can effectively identify the current technological development trend and is more suitable for the current technological innovation of multidisciplinary and cross-domain engineering.

ABSTRACT. Weak signal studies can provide decision-making support to the strategic foresight process and strategy making, and have gradually become research hotspots. To find out the developmental trajectory of weak signal studies, an intermediacy network is constructed here after collecting scientific publications from the Scopus database with the time spanning from 1975 to 2020. Different from main path analysis, more details are illustrated in the intermediacy network, from which four developmental stages can be observed. Our preliminary research indicates that the intermediacy can serve as an alternative to main path analysis on discovering developmental trajectory.

ABSTRACT. In recent years，the cross-disciplinary integration has become the main driving force of promoting scientific and technological innovation. Detecting and identifying the evolution of interdisciplinary topics thus have become increasingly significant for pushing interdisciplinary cooperation, facilitating the integration and development of disciplines, and responding to challenges in real life. This paper proposes a topic evolution analysing method using dynamic topic models in scientific articles, focusing on identifying interdisciplinary topics evolution and the corresponding subject changes. The research potentially can provide assistance in discovering emerging interdisciplinary topics and cultivating interdisciplinary talents. What’s more, it can also provide decision support for the formulation of relevant interdisciplinary policies.

ABSTRACT. The literature on innovation (eco) systems, sustainability transitions, and institutional theory can benefit from microdynamics and micro-foundational studies on innovation intermediaries. As actors who catalyze sustainability transitions in socio-technical systems by articulating new visions, demands, and expectations and acting as an impartial voice in complex networks (Matschoss and Heiskanen 2017), there is an opportunity to move beyond surveys and interview data to study the multilevel, multimodal characteristics of these entities as well as their longitudinal impact. Following a digital methods paradigm (Hutchinson 2016; Rogers 2019), this work leverages tech mining to improve our understanding of the intermediaries' socio-psychological characteristics. The Linguistic inquiry word count (LIWC ) method is therefore applied to a tweeter dataset obtained from Tweet-binder and points to how intermediaries' social media data can be linked to their "socio-psychological profile." Towards generating innovation ecosystem orchestration insights, we examine the potential of digital trace social media data text to understand how these actors engage in sustainability transition processes longitudinally ?. Preliminary results show that text can be linked to distinct personality dimensions and psychological processes. Further, our analysis will augment social network data from hyperlinks crawl to provide insights into the challenges and opportunities of using digital trace data for measuring "intangibles" such as culture and values.

ABSTRACT. As the science and technology improve rapidly, scholars have begun to explore the process of R&D and commercialization of emerging technologies. They have therefore developed an analytical view of the innovation ecosystem. This research aims to analyze the intelligent manufacturing innovation ecosystem. Applying a set of R language programs developed by this research group, we are able to do social network analysis and draw network diagrams on academic literature co-authors, patent inventors, and patentees. Data are collected from WOS Global Literature Database and Webpat Global Patent Database. Smart manufacturing, for example, can be divided into three innovation ecosystems: research, development, and application. Science metrology analysis can be applied to the global innovation ecosystem of smart manufacturing. The conclusion section will show the development process of the global smart manufacturing innovation ecosystem over the past 20 years. The research contribution can help to enhance the technological development of the smart technology innovation ecosystem, the overall knowledge and technology transfer efficiency, and to facilitate the development of emerging technology industries.

ABSTRACT. Interdisciplinary research has gradually become one of the main driving forces to promote original innovation of modern scientific research, and how to measure the interdisciplinary of science project is becoming an important topic for science fund project managements. Existing researches mainly using papers as the research object, using methods such as academic degree or institutional discipline information survey and discipline category mapping of journals to which the literature belongs. This study takes the NSF biological sciences projects and their output papers as the research object, aims to mining the interdisciplinary patterns of the field from the perspective of keywords, and capture the different or complementary characteristics of interdisciplinarity of projects. Firstly, the classification system is constructed to extract the disciplines matrix of each article reference; secondly, the matrix is tailored, and the distribution of key disciplines in each paper is classified and summarized, which is summarized into different interdisciplinary patterns, and the corresponding keywords of patterns are extracted to form the training set; finally, according to the project summary and title, the topic model is used to extract keywords set of each project to form a test set, predict the probability of each project belonging to different interdisciplinary patterns, and experts are required to evaluate and interpret the results. It will enrich the interdisciplinary measurement methods and provide references for funding policies.

ABSTRACT. Internet of things (IOT) has become a disruptive technology that turns physical objects into smart devices through existing network infrastructures. This technology also brings tons of business opportunities which is believed to create a trillion-dollar industry. The research aim of this thesis is to discover how the newly emerging IOT technology integrates into other technological applications and how business models transform in the era of IOT. This thesis will use data collected from Web of Science (Science), WebPat (Technology), and IOT-related companies (Business) to identify the technology emergence and recent business development in IOT sector. The results show that IOT has been widely integrated into a number of different fields such as communication, artificial intelligence, green energy, etc. From a business perspective, the dominant products and services of IOT companies are smart devices such as smart homes, smart vehicles and IOT-based services such as IOT platform. The research contributes to the literature on innovation studies through its rich mix of methods and its empirical focus on results of IOT technological emergences. It also contributes with comparative design. The results will be disseminated broadly with both scholarly and management audiences, and will contribute to moving global innovation forward more quickly by providing information on technology convergence through empirically studying the emergence of internet of things.

ABSTRACT. Precision medicine is defined as transforming large amounts of data into usable information through bioinformatics. Personalized medical treatment can be tailored to detect, diagnose, treat and monitor the same disease, overturning the conventional medical approach of providing consistent treatment. This study adopted a multi-level analysis of the innovation ecosystem as a research framework, and investigates How to strengthen the innovation network and the development of precision medical services through technology policy? Integrating various bibliometric, data visualization techniques and qualitative data, this study focuses on the dynamic evolution and patents of the precision medical sector over the past two decades (2000-2019). The main research question of this study is to explore how to enhance the development of precision medicine through policy tools, using the multi-level analysis of an innovative new system as the research framework. Analyzing the collaboration between actors in an innovation system provides a deeper understanding of the changes in the innovation network, and discusses how the innovation network is changing. It can be seen that which companies, research institutes or academic institutions are playing a key role in the Precision Medicine Innovation Ecosystem. Policy implication will be addressed as the concluding remark of this paper.

ABSTRACT. Blockchain is a digital database that stores records such as financial transactions which can be shared within a network in a highly secure and decentralized way. Application of Blockchain in financial transaction has been well explored in recent times and, the technology is at the center of the ongoing technological transformation. Its primary role is in decentralizing the entire network can be utilized in multiple ways to enhance day-to-day activities of the mankind and this study tries to explore the applications of blockchain on ground of the number of utility patents filed with United States Patent and Trademark Office (USPTO). As per the USPTO data, 2016 is the year when patents in this field began coming in. The number of patents show a dramatic increase in the subsequent years and we present a detailed analysis of the same through this paper. The data utilizes IPC Mapping to map the applications in a diagrammatic form to depict the evolution of Blockchain in the recent years. Network maps helps us to observe the collaborations between patent filing agencies in a systematic manner. The study reflects the evolution of Blockchain and the gradual evolution of the same in terms of application and touches upon the future implications of the process.

ABSTRACT. This research adopted multi-level analysis to analyze and explore how the AIOT innovation network and innovation ecosystem form and evolve? To research the formation of the R&D innovation network and describe the dynamic evolution of the AIoT innovation ecosystem in the past ten years（2009-2018）, We used the data of co-authorship data from Web of Science, the co-patenting data from Webpat, and also had made 37 interviews personally with enterprises or academic institutions. The research results shows that the fields of electrical and electronic engineering, computer science artificial intelligence, computer science theory and computer science information systems combine existing patent technologies and research literature that have been developed and integrate innovation across fields.National Taiwan University, National Chiao Tung University, Academia Sinica and other academic institutions, as well as Faraday Technology Co., Ltd., Phison Electronics Co., Ltd., AU Optronics Co., Ltd., Tatung Co., Ltd., Meryl Industrial Co., Ltd., Elan Electronics Co., Ltd. and other manufacturers are actively investing in the research and development of AIoT in the innovation ecosystem. The national industrial policies have shifted changed from focusing on supporting specific emerging technology industries to strengthening the infrastructure for industrial development. This research will contribute to the literature of the innovation ecosystem and innovation network. It is hoped that the research results can be provided for reference by enterprises, R&D personnel and policy makers.

ABSTRACT. Existing literature indicates that giant pharmaceutical firms which increase the research energy and share knowledges based on open innovation has gradually become the network integrator. The M&A issues have gradually emerged in recent years. Around the world, pharmaceutical firms facing the patent cliff, aging society and soaring of the medical expense. The input and output of R&D is out of proportion on many life science companies. Life sciences have encountered lots of dilemma in the recent decade. This paper attempts to conduct longitudinal study based on the 20 years of data collected from various data resources to study the changes and development trend of M&A in life science sector. The majorities of M&A activities of industries tends to horizontal mergers,but conglomerate mergers are gradually increase in recent years. In addition, the M&A of life sciences mainly focuses on the acquisition of shares （original shareholder acquisition）, and the target area is mostly domestic. The proportion of private placements has increased in recent years, and it is more active than the open market. However, most of them are individual investment. Government and other investment companies have low capital ratios, and the capital investors are scattered and most of them are neither strategic investments nor biotechnology-related industries. The findings shows the New Drug Application can be accelerated and increase the market value through different collaboration mode. In addition to attracting international capital investment, it can also promote the enterprise to increase the cooperation chances.