MIS 480: Knowledge Management Dr. Chen May 14, 2009 Kevin Prachachalerm Shantanu Soman Mike Sotelo
Table of Contents I. Introduction... 3 Advantages of SSD (Solid-state Drive)... 3 Disadvantages of SSD... 3 Trends... 4 II. Project Objective... 5 Hypotheses... 5 III. Selection of Keywords... 5 IV. Research Architecture... 6 Data Acquisition... 6 Patent Parsing and Filtering... 7 Database Engine and Framework... 7 Table Design and Data Attributes... 8 Analysis Overview... 8 Basic Bibliometric Analysis... 9 Knowledge Diffusion... 9 V. Identified Issues... 12 Structure of Spidering Program... 12 Discrepancies within the Data... 13 VI. Summary of Findings... 14 VII. Conclusion... 16 VIII. References... 17 2 P age
I. Introduction Memory may be split into two types of categories: volatile and non volatile. Volatile memory requires constant power to remain viable and loses any data the moment the system is turned off. Most types of random access memory (RAM) are included in this category. On the contrary, non volatile memory is memory that does not lose any data if a system is turned off. An example of a non volatile memory is flash memory. Due to its highly durable characteristics, flash memory technology is primarily used in multimedia memory cards and USB flash drives for the purpose of general storage or transferring any data between computer systems and compatible digital devices. This technology is also used in compact hard drives, which increases reliability in mobile applications such as laptop computers. The technology of flash memory has been around since 1980 and was developed by Dr. Fujio Masuoka of Toshiba. The name flash was suggested by a colleague of Dr. Masuoka, Mr. Shoji Ariizumi, because the erasure process of the memory contents reminded him of the flash from a camera. Advantages of SSD (Solid-state Drive) Because there are no moving parts like a conventional hard disk drive, SSDs are less fragile and more durable in nature. Along with this advantage, SSDs have significantly fast and predictable access times with no latency. Extended operating temperature range and lower power consumption makes SSDs a favorable choice for consumers when selecting a storage medium. Figure 1 Disadvantages of SSD Although there are many advantages of owning an SSD device, there are also disadvantages. The most obvious one is the average price. An SSD device may run around $500 for a 256GB storage capacity. It also has less storage capacity than conventional hard disk drives. Some SSD hard drives have a limited number of write cycles. Once this limitation has been reached by very active users, the data stored on the drive is lost. 3 P age
Trends Different forms of solid state/flash memory technology have been used in small portable devices such as phones, cameras, and other handheld devices. As more hard drive manufacturers and computer companies implement this technology into their products, SSD will become the new standard for storage devices. Figure 2 portrays an increasing trend in the development of solid state/flash memory technology with top companies like Toshiba, Mitsubishi, and Samsung leading the market. 700 600 500 400 300 200 100 0 Toshiba Mitsubishi Samsung NEC Fujitsu Hitachi Hynix Renesas Micron Sharp Figure 2 The prominent countries in the field of solid state/flash memory technology are Japan, South Korea, and the United States, with Japan having at least three times the number of patents of other countries. 900 800 700 600 500 400 300 200 100 0 Japan South Korea United States Taiwan Germany Italy Israel France Singapore Netherlands China Figure 3 4 P age
II. Project Objective The goal of this project is to conduct knowledge mapping research in the topic of Flash Memory Technology using patents collected by the USPTO database. By acquiring and analyzing information from the database, our team will be able to observe the direction of flash memory technology and identify current trends and innovation. Hypotheses Based on our data analyses, we wanted to provide valid support to three hypotheses: 1. Do the market leaders in revenue also share the status of issued patent leaders? 2. Do companies with most patents have the best lawyers? 3. Does the company with most issued patents have the most knowledgeable inventors? III. Selection of Keywords We understand that in order to obtain accurate analyses in the field of flash memory technology, it is extremely critical to select the appropriate keywords for the spidering process. From a general perspective, the obvious keyword to use was flash or memory; however, we eventually avoided these particular terms in our spidering because it would result in an excess amount of noise after the spidering was complete. With that in mind, we constantly modified our keyword based on the words found in the returning patent titles. The following set of keywords was used for the initial spidering, as shown in Figure 4: Figure 4 5 P age
IV. Research Architecture As shown in Figure 5, our research architecture consists of three main parts: Data Acquisition, Data Parsing, and Data Analysis. Figure 5: Research Architecture Data Acquisition Our team used the United States Patent and Trade Office (USPTO) website to acquire patent data. The analysis data is based on actual filed patents as opposed to issued patents. The reason for this approach is that actual patents filed capture a more precise measurement of tacit knowledge than issued patents. For example, Inventor A filing for a patent that was already issued to Inventor B may not receive approval for the patent. Although Inventor B acted earlier than Inventor A in filing the patent, this does not exclude Inventor A from possessing expert knowledge in regards to the patent. For this reason, by restricting our analyses to only issued patents, this approach would exclude a significant amount of prominent inventors who did not receive approval for a patent, which would create a bias in our findings. 6 P age
Patent Parsing and Filtering The process of patent parsing and filtering consists of three stages. 1. Spider USPTO using select keywords In the first stage, we used the initial set of keywords to perform the initial spidering of patents that were related to Solid state/flash memory technology, which resulted in 800,944 patents 2. Remove redundancy of patents In this stage, we identified all unique patents by removing any duplicating patents. This 3. Remove noise left us with 42,889 patents In the final stage, we removed further noise by using specific keywords found in the title of patents, which gave us the final count of 10,481 patents Database Engine and Framework After acquiring all data pertinent to flash memory technology, we uploaded the patents into the staging database, powered by Microsoft SQL Server. Figure 6 7 P age
Table Design and Data Attributes Based on the amount of information obtained from the USPTO repository, we organized our data using eight tables with each table having at least four attributes, as shown in Figure 7. Although we were unable to include patent claim and patent abstract due to restrictions of SQL Server, each patent still contains a significant amount of information. Figure 7: Table Design and Attributes Using the SQL database as the final repository for our filtered data, we were able to query specific patents to be used in our data analyses. Analysis Overview Bibliometric Analysis Knowledge Diffusion Patent Publication Most Issued Patents by Country Annual Issued Patents by Country Most Issued Patents by Assignee Annual Issued Patents by Assignee Most Issued Patents by Attorney Most Issued Patents by Inventor Patent Significance and Strength Most Cited Patents Patents Using Most Citations 8 P age
Basic Bibliometric Analysis In performing a bibliometric analysis on issued patents, this methodology allows our team to examine the historical development found solid state/flash memory technology and find patterns from various authors, publication, and the use of such technology. Our bibliometric analysis was performed based on patent publication of country, assignee, attorney, and inventor. All graphs and charts for patent publication may be viewed in our presentation slides. Knowledge Diffusion Figure 8: Relationship of Attorneys and Assignees using NetDraw The diagram of Figure 8 was generated using a network analysis software known as NetDraw. This visual diagram facilitated our analysis for the top companies and the attorneys they used when filing for patents. Obvious from Figure 8, McDermott, Will & Emery is the common law firm choice of Mitsubishi, while Oblon, Spivak, McClelland, Maer & Neustadt, P.C. is the primary choice for Toshiba. 9 P age
Figure 9: Citation Analysis among different countries using NetDraw For our citation analysis, we decided to focus on citations among countries to give us a better understanding of prominent countries within the field of solid state/flash memory technology. The size of the circle is determined by the amount of times the country has been cited by another. In alignment with our column graphs from the presentation slides, Japan is highly specialized in this particular field since most of the issued patents from the top assignees are Japanese companies. Being a dominant country within the field, Japan is often cited by other countries. Although South Korea was the second leader in the industry in terms of most issued patents (based on Figure 3), it is interesting to see that the United States had more patents being cited. This leads us to believe that some of the foundation or components of solid state/flash memory technology originated from the United States. We also used another network analysis software called Graphviz. In order to optimize visibility of the graph generated by Graphviz, we decided to only capture citations greater than 10, as shown in Figure 10. 10 P age
Figure 10: Citation Analysis among different countries using Graphviz 11 P age
V. Identified Issues During our data acquisition and analyses, we have encountered issues that might have been easily overlooked without careful observation. The issues that have been identified during our analyses are so critical that it would have a negative impact on our numbers. In one instance, the ranking of the top 10 companies completely changed after we had factored in these issues. Samsung was thought to be the dominant company in this field; however in actuality, Toshiba surpassed Samsung by at least 3 times as much patents. This same observation was seen throughout other parts of our analyses. Structure of Spidering Program As the keywords are important for obtaining pertinent patents, the code within the spidering program is just as crucial. As seen in Figure 11, we noticed that the given code for querying the USPTO database needed to be modified, since its original form attracted an excessive amount of noise spidering all fields as opposed to specific fields. Being that the patent claim had the most real estate for noise, we decided to avoid spidering that portion of the patent and focus on the title. Figure 11 12 P age
Discrepancies within the Data Some discrepancies were encountered during our data analyses and querying, which made our aggregate patent counts inaccurate. In an attempt to remedy these discrepancies, we prepared special queries of each case to obtain the actual number of patents. Some examples of discovered discrepancy are shown below. Figure 12: Companies with subsidiaries, errors in data entry, misspelling of company names, and different listing of company names 13 P age
VI. Summary of Findings To support the three hypotheses presented in the beginning of this documentation, the following graphs are shown below. 1. Do the market leaders in revenue also share the status of issued patent leaders? The answer is yes. The top companies from Figure 13 reflect the top companies observed in Figure 14 Samsung, Toshiba, Hynix, Micron Technology, and Sharp. Figure 13 1600 1400 1200 1000 800 600 400 200 0 1425 1263 1129 815 804 582 540 413 350 258 Total Figure 14 14 P age
2. Do companies with most issued patents have the best lawyers? Yes, most of the top companies are contracting with the best attorneys in the field. McDermott, Will & Emery seem to be the most prominent attorneys, based on Figure 15. 1200 Top Attorney with Most Issued Patents 1000 1014 800 600 400 200 0 487 401 298 281 198 178 175 155 Total Figure 15 15 P age
3. Does the company with most issued patents have the most knowledgeable inventors? Yes, based on Figure 13 inventors from Toshiba and Mitsubishi possess the most issued patents. 140 Inventors with Most Issued Patents 120 127 126 100 80 60 90 90 82 74 71 63 63 63 40 Total 20 0 Mitsubishi Toshiba Mitsubishi Mitsubishi Toshiba Mitsubishi Toshiba Mitsubishi Toshiba Toshiba Ooishi, Tsukasa Tanaka, Tomoharu Arimoto, Kazutami Hidaka, Hideto Nakamura, Hiroshi Asakura, Mikio Shirota, Riichiro Dosaka, Katsumi Iwahashi, Hiroshi Atsumi, Shigeru Figure 13 VII. Conclusion The results from the initial spidering of patents totaled 800,944. After carefully filtering and parsing out data from our staging database, the final count of patents totaled 10,481. When applying these findings to commercial use, many organizations specializing in the area of flash memory technology may gain competitive knowledge. Results may also be used to support any assumptions or speculations made by researchers within this field. Performing various analyses of the gathered data enabled our team was to determine prominent players within the field of flash memory technology. 16 P age
VIII. References 1. Convergent Semiconductors http://www.convergentsemiconductors.com/index.php?mact=news,cntnt01,detail,0&cntnt01articleid= 9&cntnt01returnid=71 2. EMS NOW http://www.emsnow.com/newsarchives/archivedetails.cfm?id=12615 3. Improved Flash Memory Grows in Popularity http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=01580375 4. isuppli http://www.isuppli.com/products/home.aspx 5. Samsung Semiconductor http://www.samsung.com/us/business/semiconductor/index.html 6. Sandisk: Timeline of Innovation, 2009 http://www.sandisk.com/corporate/timeline/years2004 2006.aspx 7. Semiconductor Timeline Notes, 2006 http://corphist.computerhistory.org/corphist/documents/doc 4803f82fa3ba8.pdf?PHPSESSID=cb4318c8e1c3e0c4931d81eb1150694f 8. Solid State Technology s WaferNEWS http://newsletters.pennnet.com/semiweekly/2083647.html 9. Toshiba www.toshiba.com 10. Wikipedia www.wikipedia.org 17 P age