Technological Route between Pioneerism and Improvement
Roberto Nani
1
, Daniele Regazzoni
2
1
Scinte s.n.c., Ranica, Italy...
Figure 1: methodology steps
The third step of the paradigm consists in fitting the IPD
data in logistic curves [1], [4], [...
10 rows shown 3784
(Below cut-off) 5,769 60.4 ...
Table 1: IPC distribution of patents of preliminary search
The final doc...
The statistical-bibliometric analysis is performed to obtain
the data needed to calculate the Intellectual Property
Densit...
3.4.1 technology I: midpoint = 1972; growth time =
14.4; saturation = 6.7
A clustering overview of patents and application...
Figure 12: 5 clusters graph of collection 1987
((support AND rotation AND tube AND extend AND press)
<in> (TITLE,ABSTRACT,...
4 CONCLUSIONS
This paper shows an experience-based way to perform a
patent search capable of gathering not only the static...
of 7

Technological Route between Pioneerism and Improvement

Published on: Mar 3, 2016
Published in: Data & Analytics      
Source: www.slideshare.net


Transcripts - Technological Route between Pioneerism and Improvement

  • 1. Technological Route between Pioneerism and Improvement Roberto Nani 1 , Daniele Regazzoni 2 1 Scinte s.n.c., Ranica, Italy 2 University of Bergamo, Dalmine, Italy Abstract This paper presents a systematic approach to determine the technological route between discovery, pioneering, radical creation and qualitative or quantitative improvement as the two extremities of a line of evolution regarding a certain inventive theme. The technological route is the result of a procedure based on a wide use of a patent search engine capable of Browsing Codes according to International Patent Classifications, and Clustering Texts to examine search results using linguistic technologies, and a tool able to analyse time data series to disclose their eventual logistic behaviour. Said procedure consists of four main steps: (i) determination of the initial search algorithm based on the characteristics of reference subject, patent or group of patents; (ii) definition and calculus of Intellectual Property Density (IPD) on the base of bibliometric results of search; (iii) IPD fitting into logistic curves and (iv) definition of a quick and semi-automated way to gather some inputs for further product innovation. The procedure still requires a minimum skill in patent searching and handling and search may have to be iterated some times before getting the desired outcome. By the way there is no need to read patents text and the time spent is comparable to a traditional search (i.e. according to EPO guidelines for search). The results obtained consist in fostering both forecasting activities by identifying quantitatively main evolution phase and problem solving activities by giving a dynamic view of the state of the art so that future problems may be taken into account in today solutions. One out of the several case studies already tested with this approach is presented with the aim of describing in detail each step performed. Keywords Patent Data Base (PDB), International Patent Classification (IPC), Intellectual Property Density IPD, logistic S-curves, emerging technologies 1 INTRODUCTION Among TRIZ users the importance given to patent resources is far behind the mere protection of R&D results. Patents represent a starting point for new inventions and a huge resource for collecting information on the way contradictions have been solved and in which different field such solutions may be adopted. Moreover the worldwide patent database contains information about the technology evolution that can be extracted so that the level of maturity of a product or process can be evaluated. The connection among patent resources and problem solving activities can be improved in order to provide the innovators with meaningful data in short times whenever a problem occurs. The amount of time needed and the unreliability of traditional patent search results cause problem solvers to under exploit patents derived data. To overcome this issue several attempts to make it better and more automated can be found in literature [3,6,7]. The aim of this paper is to differentiate from detailed and high resources time consuming approaches such as [3,7]. The novelty proposed consists in a simple and efficient experience-based approach to perform innovation driven patent investigations. 2 GOAL AND METHODOLOGY The goal of this paper is to present an organized set of steps to clearly identify the patent state of the art of a certain product or technology, so that further research activities such as technological forecasting could be performed. The main reasons why this work has been carried out are the increasing demand of technology assessment based on patent information, and the growing industrial awareness of intellectual property as an asset to foster innovation. The methodology is made up of four main steps as reported in Figure 1. The starting point is as general as possible and could simply be a request of information about the state of the art of a specific technology, a device or a function in a specific context. Once defined focus and boundaries of the matter to investigate on, the first step is performed. By browsing or searching in the patent classification the classes characterizing the matter are identified. Then a usual patent investigation is performed by crossing keywords and classes search. The procedure can be iterated several times to refine keywords on the base of the results obtained. Once the patent set is satisfying, in step 2 some statistical-bibliometric analyses are performed. In particular patents distribution and patents classification over time are plotted. Afterwards, Intellectual Property Density (IPD) has been introduced by the authors as the ratio between the cumulated number of patents filed till a specific year and the count of 4 digits classes involved till the same year. According to the authors, IPD describes quite easily the general trend characterizing relevant innovations evolution. The first patents filed when a new invention appears disclose highly innovative devices or technology that may have a wide spectrum of application and potential fields of use [8]. As the patents leading the way are a few and they can be found in several different patent classes, the IPD of an emerging technology is quite low. During technology evolution the number of patents increases and they get more and more focused on sub- systems or details that are often classified in a small number of classes. As a consequence IPD in maturity stage is much higher. There is not scientific evidence that the curve plotting IPD over time has recurrent trend, by the way the experimental case studies accomplished so far have always produced qualitatively S-shaped curves.
  • 2. Figure 1: methodology steps The third step of the paradigm consists in fitting the IPD data in logistic curves [1], [4], [5]. Data can be adequately filtered, if needed, to exclude peaks due to known reasons or to cut out the first or last figures. To do so we used a software package for analyzing logistic behaviour developed and shared by the University of Rockefeller (NY) called Loglet Lab. Loglet Lab allows to discern and analyze S-shaped curve or a succession of many S- shaped curves even if overlapped in time. If the patent set obtained in step 2 and its relative IPD trend results to be logistic, one or more curves can be found with minimal residuals. Each quantitative S-curve describes a generation of device or technology, giving precious data on mean and saturation times. The results of step 3 can be exploited for at least two different kinds of activities. If the goal of the research is to understand the trends of evolution, further observations can be done to support strategic R&D activities planning. Future events can be foreseen by identifying the super- system conditions that influenced certain evolutions path despite of others, reasoning on the resources that will run out first or applying structured tool to perform a complete technological forecasting [8]. The second way the results achieved can be exploited concerns problem solving by deriving solutions from different domains. Step four describes a singular way to perform this task. Starting form an S-curve describing IPD density behaviour we may presume that the mean year is the one in which patents filed are mostly concentrated on the specific device or technology characterizing the curve. This can be explain by the fact that, when the technology reaches its maximum growth, the interferences of the previous and following technologies is the lowest. Thus when the technologies are strictly substituting one another (non concurrent in time) the interference is negligible. This approach is used to dramatically decrease the number of patent to work on, confidently not compromising the final result. At this point a basic clustering algorithm is applied to the set of patents so that most occurring words are found and grouped in order to minimize cluster interactions. Depending on the number of patents and on their homogeneity different number of clusters can be found. Filtering clusters containing general terms is quite easy to identify one or two set of words characterizing the novelty. Then a new patent search is done using such words as keywords without imposing any class constraint. The result gathered is considered in terms of patent class to identify commonalities with the starting matter, in terms of problem to address or solutions found. The fourth step is quite complex but completely automated and its definition is derived by practical application. Actually, it is not time consuming and the advice achieved mostly lead to valuable results. The following part of the paper is dedicated to the application of the steps described so far in order to give detailed guidelines to perform innovation oriented patent investigation. 3 METHODOLOGY APPLICATION The following paragraphs report the application of the four step methodology to a real case study concerning textile looms. In particular the focus is put on the weft insertion technology of a weaving machine, which has the overall function of gripping a fibre and moving it along a specific path. 3.1 Step 1: preliminary patent search To search for patents describing the state of the art of weft insertion technology the following Boolean algorithm has been used with a patent database [10]: ((gripper*) <in> (TITLE,ABSTRACT,CLAIMS) ) AND ((weft* OR thread* OR filament* <in> (TITLE, ABSTRACT, CLAIMS))) (1) returning about 6.500 documents of European, American and International patents and applications: Collections searched: European (Applications - Full text), European (Granted - Full text), US (Granted - Full text), WIPO PCT Publications (Full text), US (Applications - Full text) 6,458 matches found of 11,531,757 patents searched distributed according to International Patent Classification (IPC) in the following classes: IPC-R Code- 4 digit Items % Bar Chart D03D D — Textiles; Paper; Weaving; Woven 776 8.1 % B65H B — Performing Operations; Transporting; Conveying 614 6.4 % B65B B — Performing Operations; Transporting; Conveying 448 4.6 % B25J B — Performing Operations; Transporting; Hand Tools 399 4.1 % B41F B — Performing Operations; Transporting; Printing 312 3.2 % A61B A — Human Necessities; Medical or Veterinary Science; H 283 2.9 % B65G B — Performing Operations; Transporting; Conveying 262 2.7 % E21B E — Fixed Constructions; Earth or Rock Drilling; Mining 245 2.5 % B25B B — Performing Operations; Transporting; Hand Tools 233 2.4 % B29C B — Performing Operations; Transporting; Working of PLA 212 2.2 %
  • 3. 10 rows shown 3784 (Below cut-off) 5,769 60.4 ... Table 1: IPC distribution of patents of preliminary search The final documental collection of the “filamentary manipulation (gripping)” is obtained by the following Boolean query that crosses the function “to grip” with the most representative classes: (((d03d OR b65h) <in> IC ) AND ((gripp*) <in> (TITLE,ABSTRACT,CLAIMS))) (2) Collections searched: European (Applications - Full text), European (Granted - Full text), US (Granted - Full text), WIPO PCT Publications (Full text), US (Applications - Full text) 8,700 matches found of 11,531,757 patents searched 3.2 Step 2: patents analysis and IPD determination According to the results obtained with the algorithm (2) the “filamentary manipulation (gripping)” starts its technological path on 1932 with a “gripper operating mechanism” (Figure 2). From 1932 to 2008 the knowledge of “filamentary manipulation (gripping)” characterizes filed patents and applications capturing electric, mechanic, chemic resources from the technological branches. Figure 2: first patent considered concerning filamentary manipulation (gripping) From the first patent to the last completed year data are collected and plotted in order to highlight: - number of patents filed per year (Figure 3); - number of 4 digits IPC classes in which patents are classified per year (Figure 4); - Cumulative number of patents filed per year (Figure 5); - Total count of classes in which patents have been classified per year (Figure 6). 0 50 100 150 200 250 300 350 400 450 Filed Year 1958 1966 1969 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 Figure 3: Number of patents per year 0 10 20 30 40 50 60 70 80 90 100 Filed Year 1958 1966 1969 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 Figure 4: Number of involved IPC classes per year 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 Filed Year 1958 1966 1969 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 Figure 5: Cumulative number of patents filed per year 0 50 100 150 200 250 300 350 Filed Year 1930 1970 1975 1980 1985 1990 1995 2000 2005 2008 Figure 6: Total count of classes per year
  • 4. The statistical-bibliometric analysis is performed to obtain the data needed to calculate the Intellectual Property Density (IPD) number as the ratio between the cumulative number of patents and the classes involved (i.e. the ratio among the figures plotted in Figure 5 and Figure 6). The curve describing IPD behaviour over time is shown in Figure 7. 0 5 10 15 20 25 30 Filed Year 1930 1965 1970 1975 1980 1985 1990 1995 2000 2005 2008 Figure 7: Intellectual Property Density IPD 3.3 Step 3: logistic curves fitting This step is the most important as it provides the empirical evidence that IPD trend is a logistic trend. By using Loglet Lab software we found out that the distribution of Figure 7 is actually fitting with minimal residuals (3% maximum) to a set of three S-curves. Figure 8-10 show respectively the S-shaped curves, the bell curves and the Fisher-Pry transform curves. Figure 8: S-shaped curves fitting IPD data Figure 9: Bell-shaped curves fitting IPD data Figure 10: Fisher-Pry transform curves fitting IPD data From the fitting of the curves three main technologies concerning “filamentary manipulation (gripping)” emerged. Each curve is characterized by the following data: • Technology I: midpoint = 1972; growth time = 14.4; saturation = 6.7; • Technology II: midpoint = 1987; growth time = 11.7; saturation = 9.8; • Technology III: midpoint = 1999; growth time = 17.3; saturation = 12.4. The technologies defined in this way have to be interpreted according to breakthrough technological changes regarding, for examples, new materials, new chemical-physical discoveries/principles or due to socio- political events, new standards. On the next paragraph a clustering algorithm is applied to applications and patents of said saturation points: (((d03d OR b65h) <in> IC ) AND ((gripp*) <in> (TITLE,ABSTRACT,CLAIMS))) (2) Filed Year Items % Bar Chart 1999 305 3.5 % 1987 320 3.7 % 1972 112 1.3 % Table 2: number of patents 3.4 Step 4: new IPC classes of interest In the next paragraphs 3.4.1, 3.4.2 and 3.4.3 a clustering algorithm is applied to patents and applications of specific saturation points: 1972, 1987 and 1999. Analyzing the text of patents and applications, terms are grouped in order to obtain the minimum number of cluster having almost no cross connections. The result can be automatically shown by a graph. Each cluster is composed by a number of terms, describing both devices and actions quite homogeneous for meaning and field. A quick scan of the terms is enough to associate TRIZ Inventive Principles to each cluster. By the way there no need to go through all the clusters if in the working context a specific goal is defined.
  • 5. 3.4.1 technology I: midpoint = 1972; growth time = 14.4; saturation = 6.7 A clustering overview of patents and applications filed on 1972 considers specific devices and actions constituting the Boolean algorithm (4): Cluster Overview 5 Clusters For work file: 1972 (112 items ) Cluster Descriptive words 1 comprise, loom, roller, provide, apparatus, mean, move, device, control, include 2 sheet, mechanism, gripper, mount, press, transfer, conveyor, include, movement, printing 3 form, engage, surface, end, material, drive, provide, apparatus, lower, length 4 mean, member, parallel, relative, grip, pair, include, release, pass, groove 5 release, side, arrange, adjacent, movement, locate, operation, hold, first, time Table 3: clustering overview of collection 1972 Figure 11: 5 clusters graph of collection 1972 ((release AND groove AND arrange AND movement) <in> (TITLE,ABSTRACT,CLAIMS)) (4) Collections searched: European (Applications - Full text), European (Granted - Full text), US (Granted - Full text), WIPO PCT Publications (Full text), US (Applications - Full text) 6,400 matches found of 11,374,720 patents searched The Boolean algorithm (4) produces a collection of about 6.400 documents, the IPC resource being the technological branch (IPC): - F16L 37/00 Couplings of the quick-acting type ; - A61M 5/00 Devices for bringing media into the body in a subcutaneous, intra-vascular or intramuscular way; Accessories therefor, e.g. filling or cleaning devices, arm rests . 3.4.2 technology II: midpoint = 1987; growth time = 11.7; saturation = 9.8 A clustering overview of patents and applications filed on 1987 considers specific devices and actions constituting the Boolean algorithm (5): Cluster Overview 10 Clusters For work file: 1987 (320 items ) Cluster Descriptive words 1 mount, support, include, horizontal, comprise, use, adjacent, rotation, engage, recess 2 apply, stack, apparatus, grip, relate, say, method, improve, feeding, position 3 roll, web, direction, material, contact, winding, device, determine, include, part 4 sheet, apart, grip, spread, example, transport, location, first, move, apparatus 5 arrange, fold, gripper, printed product, apparatus, advance, first, extend, needle, invention 6 clamp, weft, thread, side, element, move, shed, weft thread, loom, end 7 mount, provide, shaft, include, cam, position, drive, dispose, introduce, end 8 support, surface, rotation, point, outer, tube, extend, end, form, press 9 move, object, stack, conveyor, deliver, include, grip, release, vacuum, carry 10 process, transfer, arm, transport, perform, mean, position, control, path, sheet Table 4: clustering overview of collection 1987
  • 6. Figure 12: 5 clusters graph of collection 1987 ((support AND rotation AND tube AND extend AND press) <in> (TITLE,ABSTRACT,CLAIMS)) (5) Collections searched: European (Applications - Full text), European (Granted - Full text), US (Granted - Full text), WIPO PCT Publications (Full text), US (Applications - Full text) 1,210 matches found of 11,374,720 patents searched The Boolean algorithm (5) produces a collection of about 1.200 documents, the IPC resource being the technological branch (IPC): - B29C SHAPING OR JOINING OF PLASTICS; SHAPING OF SUBSTANCES IN A PLASTIC STATE, IN GENERAL; AFTER- TREATMENT OF THE SHAPED PRODUCTS, e.g. REPAIRING. 3.4.3 technology III: midpoint = 1999; growth time = 17.3; saturation = 12.4 A clustering overview of patents and applications filed on 1999 considers specific devices and actions constituting the Boolean algorithm (6): Cluster Overview 5 Clusters For work file: 1999 (305 items ) Cluster Descriptive words 1 sheet, device, form, grip, machine, method, station, gripper, position, comprise 2 invention, relate, say, gripper, comprise, define, provide, include, section, guide 3 mean, apparatus, control, arrange, transfer, provide, convey, product, form, first 4 include, move, position, first, grip, stack, engage, apparatus, rotation, hold 5 end, support, portion, reel, rotatably, comprise, mount, shaft, arrange, provide Table 5: clustering overview of collection 1999 Figure 13: 5 clusters graph of collection 1999 ((include AND move AND position AND engage AND rotation AND hold) <in> (TITLE,ABSTRACT,CLAIMS)) (6) Collections searched: European (Applications - Full text), European (Granted - Full text), US (Granted - Full text), WIPO PCT Publications (Full text), US (Applications - Full text) 21,483 matches found of 11,374,720 patents searched The Boolean algorithm (6) produces a collection of about 21000 documents, the IPC resource being the technological branch (IPC): - G11B 17/00 Guiding record carriers not specifically of filamentary or web form, or of supports therefore.
  • 7. 4 CONCLUSIONS This paper shows an experience-based way to perform a patent search capable of gathering not only the static state of the art of a specific technological domain, but providing quantitative data to rely on to perform problem solving as well as technological forecasting. The methodology is rooted in a simple but unconventional use of patent search engine and of a software package for logistic curve analysis available for free. The most relevant result achieved within the experience behind this paper is that the ratio among number of patents and their distribution on IPC classes, called Intellectual Property Density, behaves in a logistic way. This assumption is based on a number of case studies differing for field of applications and maturity of technology in which the IPD trend has always shown such characteristic. After the evolution steps of a device or a technology have been defined and each of them is associated to a quantitative S-shaped curve, both problem solving and technological forecasting can be performed with relevant benefits. Actually the maturity level of present state of the art systems is assessed and even maturity and decay period can be foreseen and taken into account to search for solutions exploiting resources that won’t run out in the near future or already thought to overcome problems that still have to appear. Concerning the specific application reported in this paper the emerging technological collections defines new resources that are practically unknown from prior art analysis by Boolean algorithm (2), as shown on table 6: • F16L 37/00 Couplings of the quick-acting type; • A61M 5/00 Devices for bringing media into the body in a subcutaneous, intra-vascular or intramuscular way; Accessories therefor, e.g. filling or cleaning devices, arm rests ; • B29C SHAPING OR JOINING OF PLASTICS; SHAPING OF SUBSTANCES IN A PLASTIC STATE, IN GENERAL; AFTER- TREATMENT OF THE SHAPED PRODUCTS, e.g. REPAIRING; • G11B 17/00 Guiding record carriers not specifically of filamentary or web form, or of supports therefore. IPC-R Code- 4 digit Items % Bar Chart B65H B — Performing Operations; Transporting; Conveying 6375 45.0 % D03D D — Textiles; Paper; Weaving; Woven 1140 8.0 % B41F B — Performing Operations; Transporting; Printing 853 6.0 % ….. ….. …. ….. B29C B — Performing Operations; Transporting; Working of PLA 184 1.3 % …. …. …. … G11B G — Physics; Information Storage; Information Storage B 63 0.4 % …. ….. …. … F16L F — Mechanical 21 0.1 % Engineering; Lighting; Heating … …. …. …. A61M A — Human Necessities; Medical or Veterinary Science; H 8 0.0 % Table 6: confront of collocation of new IPC resources respect to main IPC classes of prior art Boolean algorithm (2) REFERENCES [1] Gibson N., (1999) The Determination of the Technological Maturity of Ultrasonic Welding, The TRIZ Journal, http://www.triz- journal.com/archives/1999/07/a/index.htm [2] N. Leon, J. Martinez, C. Castillo, (2005) Methodology for the Evaluation of the Innovation Level of Products and Processes, proceedings of TRIZCON05, Brighton MI USA, April 2005 [3] Cascini G., Neri F., "Natural Language Processing for patents analysis and classification", Proceedings of the TRIZ Future 4th World Conference, Florence, 3-5 November 2004, published by Firenze University Press, ISBN 88-8453-221-3. [4] Kucharavi, D., De Guio R., Technological Forecasting Assessment of Barriers of Emerging Technology IAMOT 2008, p.20, Dubai, UAE, 2008 [5] M. S. Slocum, C. O. Lundberg, (2007) Case Study: Using TRIZ to Forecast Technology, The Triz Journal, http://www.triz- journal.com/content/c070507a.asp [6] Han Tong Loh, Cong He and Lixiang Shen, (2006) Automatic classification of patent documents for TRIZ users World Patent Information, Vol. 28, Issue 1, March 2006, Pages 6-13 [7] Han Tong Loh, Cong He (2008) Grouping of TRIZ Inventive Principles to facilitate automatic patent classification Source, Expert Systems with Applications: An International Journal, Volume 34 , Issue 1 (January 2008), ISSN:0957-4174 [8] V. Souchkov (2007) Differentiating Among the Five Levels of Solutions, The Triz Journal, http://www.triz- journal.com/archives/2007/07/02/ [9] www.delphion.com CONTACT Roberto Nani Scinte s.n.c. 24020, Ranica (BG), Italy E-mail: info@scinte.com Phone: +39 (035) 513683 FAX: +39 (035) 513683 Daniele Regazzoni University of Bergamo, 24044, Dalmine (BG), Italy E-mail: daniele.regazzoni@unibg.it Phone: +39 (035) 2052353 FAX: +39 (035) 2052077

Related Documents