Adoption of the so-called pro-patent policy in Japan is currently lagging twenty years behind that in the U.S. Japanese companies are now trying to catch up by attempting to maximize their use of available patent information. The latest information analysis tools present patent information in such a way as to facilitate evaluation of the originality, technological importance, and industrial applicability of inventions. Also, such tools allow prediction of new technological areas, in relation to which patents applications may be filed in the near future, so as to support development strategy planning as well as investment strategy planning.

Among such tools, XLUS (pronounced kailas), produced by SO-TI, Inc., a Japanese venture enterprise, is attracting great attention. XLUS creates an “information macrocosm scattered with inventions.” “Stars” that represent inventions make up local “star systems” and “galaxies.” In this information macrocosm, stars owned by certain companies, star systems which companies are trying to control, and competitions in individual galaxies can be grasped.

In June 2008, SO-TI, Inc. started providing services in the U.S.; most notably, the U.S. patent analyzing ASP services “XLUS Eagle G” and “XLUS Eagle W.” SO-TI gives members of American law firms, companies, research institutes and universities an opportunity to journey to the “invention macrocosm.” A new window through which patent searches and analyses can be performed has now been opened by SO-TI.

The technology of SO-TI has been developed by two outstanding researchers. Dr. Nakamura, the CEO of SO-TI, specializes in operations research, intellectual property analyses and data mining. Dr. Takaeda, the CTO of SO-TI, specializes in polymer statistical mechanics, study of polymer dilute solution properties and material informatics. Waseda University, Mitsubishi Corporation, and several venture capital companies have invested in SO-TI which was founded in 2006. The ASP services “XLUS Green” and “XLUS White”, which handle Japanese published patent applications, have been available since 2007.

Dr. Nakamura says, “By viewing a vast array of information displayed in a panoramic manner, the connections among individual pieces of information, which may have been previously difficult to comprehend, can be easily understood. This approach is important for encouraging innovation in technological fields and for aiding planning of a patent acquisition strategy. One of our missions is to provide a tool capable of performing high-dimensional analyses that are far beyond the analytical capabilities of human beings.”

Using his computing expertise, Dr. Takaeda was able to realize the concept envisioned by Dr. Nakamura. He commented that, “Information on Japanese patents concerns approximately six million published applications and the number of applications is increasing by four hundred thousand per year. Parallel processing technology can be used to analyze such an enormous amount of data in a short time; such an approach has already been adopted in genomic analyses in bioinformatics.”

In Japan, a number of major electronics companies and automobile manufacturers have already installed XLUS. In addition, although marketing activities in the U.S. have just got underway, inquiries from various government agencies are already being received.

---

Hurdles in Patent Searches and Analyses

How have patent searches and analyses been carried out by patent analysts to date? There are three main hurdles to be tackled to accomplish these tasks.

First of all, the number of patents that a patent analyst has to search through and analyze is constantly increasing. The U.S. patent and trademark office (USPTO) is inundated with enormous numbers of new applications every year. Newly published applications and issued patents are stored in the USPTO database on a daily basis. A patent analyst is forced to try and deal with this exponential explosion of available information.

Secondly, conventional searching approaches require sophisticated search skills as well as knowledge of the technology to be searched. Using search engines, hundreds of thousands of patents can be roughly searched. An experienced analyst working for a Japanese manufacturing company says, “A patent analyst has to refine the obtained search results using search criteria (search formulas) so as to make a short list of patents and applications to be subjected to thorough review. It is always challenging for a patent analyst to deal with the latest technology and cross-sectional technologies, for which proper keywords and patent classifications for narrowing down the search do not yet exist.” Dr Nakamura points out, “Without search skills and technological knowledge, results of a satisfactory level cannot be obtained. Therefore, the patent search and analysis field has become almost completely dominated by people who specialize in these tasks and are not necessarily involved in strategy planning and investment decision making.”

Thirdly, to date, patent search reports have often not been sufficiently understandable to clients. An example of such a search report is one that lists the patents together with analytical comments. Usually, clients are not happy if the number of patents listed is too large. On the other hand, if the report only lists a small number of patents, the credibility of the report might be questioned by the client. A client-oriented way of communicating that is comprehensible and credible is now required.

---

New Approaches Realized by XLUS

Compatible with information explosion

In order to search for relevant patents, XLUS utilizes a scheme called “tf-idf.” Using this scheme, the similarity between documents is calculated. Unlike in a conventional keyword search, a user may input natural language sentences as a query. The sentences thus input are first broken down into words and the importance of each word is calculated on the basis of the frequency of appearance. XLUS then retrieves patents having a high level of similarity with the query. The recent introduction of parallel computing technology has enabled XLUS to perform a large number of calculations within a few seconds. Dr. Takaeda explains, “For instance, more than forty million words including compound terms appearing in all United States Patents are utilized in the retrieval. The accuracy of the search is ensured even if the number of patents to be retrieved continues to increase.”

---

No special skills required

Figure 1 shows an interface screen of “XLUS Eagle G.” Any idea may be input into the query box in the form of natural language sentences. Alternatively, a user may copy and paste text data from other sources including academic papers and websites of other companies. One hundred relevant patents will be retrieved immediately. As shown in Figure 2, some of the thus-retrieved patents, which seem to be particularly relevant, can be set as a new query, so that the full text of those patents is used to refine the search. In this manner, XLUS does not necessarily require a user to prepare a search formula. Therefore, even without much knowledge of the technology to be searched, retrievals and screening analyses can be conducted.

---

Can be used by anyone

This may lead to a revolution in utilization of patent information. In addition to researchers and patent analysts, people who are in charge of planning management strategy or marketing strategy can become potential users of such information. Furthermore, analysts in financial institutes and investment companies can take such information into account when conducting enterprise valuations and product/service evaluations. Since the relationships among pieces of information, which had previously been invisible, can now be visualized, more accurate evaluations can be made.

---

Intuitively understandable - Radar chart

One of the core technologies of SO-TI resides in the original way in which search results are presented. In order to illustrate the similarity and technological relationships among patents, XLUS produces a radar chart in addition to an ordinal patent list. Figure 3 shows a radar chart produced by “XLUS Eagle G,” in which each dot indicates one patent. Figure 4 shows a large scale radar chart prepared by “XLUS Eagle W,” in which each dot is a cluster made up of one or more patents that are particularly similar. The distances between dots indicate the degree of similarity. Such radar charts are drawn by taking into account the full texts of the query and retrieved patents. Conventionally, search engines have indicated the similarity of documents only relative to the query. In an example shown in Figure 5, three similarity relationships between the query and respective documents are considered. As shown in Figure 6, on the other hand, six similarity relationships can be indicated by considering all of the words used in the respective documents. The radar charts of XLUS allow a user to obtain a panoramic view of retrieved patents and provide extensive information that is intuitively understandable.

---

Powerful tool for patent landscape search and state-of-the-art search

It takes much time and effort to conduct a patent landscape search and a state-of-the-art search, which are important for determining the direction of research activities. Most of the time is spent reading the retrieved documents to categorize them. Such human-based categorization tends to be subjective, and the output varies depending on the skill of the individual. In contrast, “XLUS Eagle W” electronically categorizes thousands of patents into clusters and places clusters that are similar to each other in the same area (see Fig. 7). Thus, an objective output can be obtained very quickly.

A single radar chart reveals the development trends and strategies of competitors; and the technological strengths and weaknesses of each company. Through the radar chart, activities of individual companies can be visually and intuitively grasped. XLUS enables a strategist to consider possible future moves like a chess player.

---

Marketing activities started

SO-TI is planning to open a U.S. office in 2008. Right now, they are performing market research. Dr. Nakamura suggests, “It has been noted that the mindset of intellectual property management divisions in the U.S. is quite different from that in Japan. For example, Japanese IP management divisions tend to be interested in the area of their own company’s technology. Sometimes, they are relieved when they learn that other companies are also involved in their area of focus, since they feel it confirms that they are on the right track. In contrast, U.S. companies are more interested in the creativity of their own inventors.”

One U.S. patent attorney suggests, “XLUS has great potential for improving search productivity. Usually, searches are outsourced to search firms. With XLUS, patent law firms can internally conduct some of such search duties. Also, I feel that search firms in the U.S. should install XLUS and heavily utilize it.”

Dr. Nakamura further points out, “In the U.S., financial institutes and investment companies have expressed great interest in XLUS. Utilization of our patent analysis methods and their expertise can encourage early investment in emerging technologies.”

In addition to changing the style of existing patent searches, XLUS is now triggering innovation in the strategy management operations of U.S. companies. The progress of XLUS and SO-TI, Inc. will continue to be watched with great interest.

START PAGE