--To the press -
The world's first artificial intelligence application technology to predict information leaks, etc.
Developed email audit product "Lit i View EMAIL AUDITOR"
UBIC Co., Ltd.
Masahiro Morimoto, President and CEO
2-12-23 Konan, Minato-ku, Tokyo
(TSE Mothers Code Number: 2158)
(NASDAQ ticker symbol: UBIC)
UBIC (Headquarters: Minato-ku, Tokyo, President Masahiro Morimoto), which is listed on the TSE Mothers and Nasdak in the United States, has announced that it is a "hotbed" for employees' business emails such as cartel and information leakage. We have made it possible to perform constant auditing, which was considered impossible with conventional mail auditing technology because it takes too much man-hours.Developed a new software "Lit i View EMAIL AUDITOR" that artificial intelligence automatically learns auditing methods and supports auditing by making full use of Knowledge (knowledge / knowledge). The service will be provided from April 4st.
The service does not start auditing emails after something goes wrong, but it catches emails that make you feel "predictive" on a regular basis in a timely manner, which prevents fraud in various companies. It has the effect of preventing.It is the best solution in terms of corporate compliance measures, and it is an unprecedented service not only in Japan but also in the world.
The service usage fee basically depends on the number of e-mail accounts, and the annual contract is about XNUMX yen per audited employee.At this point, we believe that most domestic manufacturing companies and companies expanding overseas have a need to introduce this service.
Why are e-mail predictive audits required now?
In recent years, it has been reported that Japanese companies have been ordered to pay a large surcharge for being involved in price cartels for automobile parts in the United States and Europe, and in the worst case, the employee in charge is imprisoned.
These were determined to violate the US antitrust law, which is Japan's antitrust law, and are a major barrier for Japanese companies to expand their business globally.At the same time, there is an increased risk that technical information of companies, which can be said to be the lifeline of the Japanese economy, will be leaked to the outside during the process of disclosure of evidence in administrative investigations, which has greatly reduced the credibility of Japanese companies as well as their profits.
So how can we prevent such corporate fraud?The first thing everyone thinks about is to check employee business emails on a daily basis to understand in advance what could lead to fraud or crime and to prevent it.However, it is just "easy to say and difficult to do", and the current situation is that very few companies actually take such measures.
This is because there is a limit to the extraction work with functions such as "keyword search" using conventional software, and in order to increase the capture rate, it is necessary to reset keywords or manually check.It was a heavy burden for companies, and considering "cost-effectiveness", I had to give up predictive auditing in the end.
Based on these circumstances, we have been working on the development of an email auditing system that is most effective at low cost from a completely new approach that sets it apart from conventional email auditing tools, and launched "Lit i View EMAIL AUDITOR". I completed it.The key concepts are "behavioral information science," "artificial intelligence application technology," and "automatic learning."Then, based on these three concepts, the features of the service will be explained below.
A new way to read big data brought about by "behavioral information science"
In this service, the text of the e-mail is decomposed to the part of speech level, and what kind of words and combinations are used among the target people are analyzed.In the past, analysis such as searching by sending / receiving date / time, sender / receiver name, title, etc. was only possible, but a new high-dimensional email content analysis that "understands the action content and visualizes important things" Is now possible.Furthermore, by using a two-layer correlation analysis function that includes not only individual units but also organizational units, it is possible to identify key persons within inter-organizational and exchange relationships, and analyze exchanges between hidden individuals and companies. I can do it.
These functions are the result of our unique approach to "behavioral information science."In other words, it is a fusion of "information science" (statistics, mathematics, data mining, etc.) and "behavioral science" (psychology, criminology, sociology, etc.), and analyzes what the conventional approach has happened in the past. In contrast to "finding facts", behavioral information science enables "analysis of human behavior and thoughts" by discovering risks more efficiently and effectively by reading human behavior and communication. I will.
For example, we are exploring possible problems in the future, such as what kind of behavior employees will take from what background, what kind of community they belong to, and what kind of behavior they will take.
The "strength" of our development process is also the "uniqueness" of the member composition of the project team.The chief executive officer of research and development has a "philosophy" background, is an expert in software development, and is the inventor of the concept of behavioral information science.In addition, under him, a staff member who has experience in researching quantum mechanical phenomena as a doctor of physics and a researcher at a specialized institution built the main logic of the data analysis function.In addition, a doctor of computational linguistics who is familiar with Japanese, Korean, English, Danish and four Japanese and European languages is in charge of linguistic analysis."Behavioral information science" is only possible when experts from different fields come together.
? Weighting by "artificial intelligence application technology"
As a pioneer of comprehensive discovery support companies in Asia, we have dealt with many projects of leading Japanese companies, and the number of projects has exceeded 355 (as of January 2014).The vast amount of data currently owned by UBIC has reached 1 terabytes.From the results of these analyzes and research, we have succeeded in developing our own artificial intelligence application technology "Predictive Coding ®" that can learn the investigator's research method and automatically extract related emails.
Until now, many lawyers have spent a lot of money and time visually reviewing e-mails such as e-discovery.However, unlike the age of paper, the amount of information contained in electronic data is orders of magnitude larger. Converting the information stored on one personal computer into paper is equivalent to four 2-ton trucks.Predictive coding teaches artificial intelligence the patterns of judgment and testing that veteran lawyers and inspectors use to find evidence, leaving most of the analysis work to the computer.Data analysis speed is about 4 times faster than skilled lawyers, and its accuracy exceeds 4,000%.As a result, it is possible to reduce the cost to about 90/10 of the conventional one.
Currently, only a few companies in the world, including us, are developing predictive coding from algorithms and analyzing data, and we are the only company that fully supports Asian languages.
With conventional services, only email extraction using technology based on keyword search is possible, and there are adverse effects such as excessive email extraction by setting a large number of keywords and emails that do not hit keywords cannot be extracted. The service can automatically extract the emails that the auditor wants with high accuracy.
? "Automatic learning" further improves extraction accuracy
Normally, the extraction settings for auditing cannot be set to 1% at one time.Therefore, daily maintenance is very important, such as setting new keywords that appear in the audited email content and deleting keywords that extract emails excessively.Our automatic learning function allows the program to automatically capture audited emails as related data and improve extraction accuracy.
Based on our experience in international litigation support and fraud investigation, we have prior knowledge of specialized fields that could not be supplemented by conventional IT and statistics alone, and we have an artificial intelligence "Virtual Data Sie".We have succeeded in developing "Ntist".A "virtual data scientist" who has learned the inspector's investigation method will automatically sort emails and accurately extract related emails.
In addition, the keywords to be updated are automatically learned and applied to subsequent audits, so the extraction accuracy improves daily. "Lit i View EMAIL AUDITOR" is equipped with a dictionary analyzed and extracted by "Virtual Data Scientist" that can be applied to various audit cases by utilizing UBIC's knowledge so far.Since the dictionary is automatically updated, the accuracy is improved every day, and the work cost and man-hours required for conventional email audit tools, such as enormous keyword setting and reduction of unnecessary browsing, are greatly reduced.
Until now, it has been an audit response to archive all business emails of employees.However, from now on, companies will be required to go further and reduce the risk of fraud and proceedings through timely and appropriate email audits. It can be said that the "era of email compliance" has arrived.
The ultimate goal of email auditing is not to monitor employees. "Protecting important employees and clients".With our solutions, we're free from fraudulent soil and allow employees and businesses to focus more on their production activities than ever before.We believe that such deterrent effects will greatly contribute to compliance measures, which in turn will lead to the improvement of corporate social credibility.
About UBIC
President Masahiro Morimoto
Address 2-12-23 Konan, Minato-ku, Tokyo Meishan Takahama Building
UBIC Co., Ltd. is an e-discovery business that preserves, investigates, and analyzes electronic data required in international cartel investigations, investigations related to the Federal Foreign Anti-Corruption Act (FCPA), IP proceedings, and PL proceedings. In addition to the electronic discovery support project), a behavioral information data analysis company that provides computer forensic investigation services that conduct investigations centered on electronic data.We have the world's highest level of technology for Asian language and the laboratory with the largest processing capacity in Asia. December 2007 Established a US subsidiary.Provided legal proceedings related to Asian companies from both Asia and the United States. At the end of 12, we developed our own eDiscovery support system "Lit i View ®" that enables eDiscovery in international litigation even within companies, and from October 2009, we developed "UBIC" as a cloud service. Started providing "Legal Cloud Service".In March 2011, we developed the world's first "Predictive Coding®" technology for Asian languages and succeeded in putting it into practical use.
Established on August 2003, 8. Listed on TSE Mothers on June 8, 2007. Listed on NASDAQ on May 6, 26.The capital is 2013 yen (as of March 5, 16).