Academic journal article
By Porter, Alan L.
Research-Technology Management , Vol. 50, No. 2
Reflect back briefly on what was considered good manufacturing management circa 1960 and what we expect today. Prior to the quality movement, manufacturing operations were based heavily on intuition. Experienced shop floor personnel knew their machines and generated a certain level of efficiency and quality-few suspected that anything better was possible. Then came the quality movement-collecting performance data along with statistical analyses enabled supervisors to discern much finer performance tolerances. And, lo and behold, production processes improved dramatically, leading to far fewer flaws in automobiles, integrated circuits or whatever. And that provided tremendous competitive advantage to the Japanese and others who implemented successful quality management (1).
Managers in most fields are exploiting data to enhance performance. Finance, logistics, marketing, and sales depend heavily upon empirically based knowledge. Even sports management is being made over. The headline in a Sporting News article reads: "Playing the numbers game: now more than ever, the sports world is looking to statistics for performance-enhancing insight, fueling the quest to devise perfect predictors of success" (2). The article discusses the expanding use of empirical indicators across all the major sports. This notion gained popular notoriety with Moneyball-the tale of how Billy Beane, as general manager of baseball's Oakland A's, gets more bang for his limited bucks by using a variety of telling player statistics (3). For instance, CERA-catcher's earned run average-incorporates a catcher's talent for handling pitchers together with other defensive skills. Data convey valuable intelligence.
How about R&D management? I assert that research managers still reach decisions based largely on intuition. For example, the cornerstone in allocating federal research funding by the National Science Foundation and the National Institutes of Health is peer review. When the government confronts difficult science and technology ("S&T") issues, it turns to the National Academies (Science, Engineering, and the Institute of Medicine) for eminent expertise. This use of expert opinion is certainly laudable, particularly when compared to the alternative of uninformed, arbitrary choices. But, the point of this article is that we can do much better by incorporating a richer base of empirical information into our R&D management processes. The data and the tools to analyze them are available now. It's time to augment expertise with empirical knowledge. In so doing, we should be able to enhance our effectiveness at least as much as has been proven possible in manufacturing and other arenas.
I see payoffs in R&D management from the introduction of carefully targeted mining of R&D information resources. These range over the spectrum of R&D and innovation processes. The researcher contemplating a new project can position this work against what others are doing to identify novel approaches, locate potential collaborators, and exploit the most promising opportunities. The portfolio manager can optimize resource allocation to target the timeliest opportunities and take full advantage of his or her assets. The intellectual property manager, new product developer, and other technology managers can similarly enhance their positioning and payoffs (4). Managers who comprehend the research landscape, backed up by explicit information, have a major competitive advantage over counterparts driving solely on intuition. Let's consider how exploiting R&D information resources can make a difference now.
Tech Mining Concepts
The analogy to production and sports management understates the challenge. Empirical knowledge in those domains derives mainly from numerical data. For S&T, the data are a mix of numerical and text. "Tech Mining" is my shorthand label for text mining of technical information resources (5). It depends upon an understanding of what matters in technological innovation processes, text mining tools to get at the pertinent intelligence, and effective delivery of findings to users. …