Taking the unique technology cultivated in the litigation support field, UBIC now utilizes behavior informatics to analyze big data and other information collections.

What is Behavior Informatics?

As an approach for your big data analysis, we propose our original idea called Behavior Informatics.
Since we see big data as a universe of human thought and behavior,
we believe our concept facilitates a more efficient retrieval of information from big data.
Indeed, we seek to uncover the humans hidden in the data.

Why does UBIC apply the concept of Behavior Informatics to its big data analysis?

We found out this approach works well through our experience of engineering software for international litigation support over the years. We were seeking a possible solution that will enable us to meet the demand of classifying business documents―emails, text data, contracts, blueprints, and reports―under a tight schedule.

Even though it seemed like a daunting task to develop a system that can reflect the decision making tasks of an expert, in this case, an experienced attorney, let alone process a tremendous amount of unstructured data which continues to grow rapidly as business activities generate them daily in forms of emails and business documents, we decided to work on this.

Thus, we set out to develop a fundamental technology called Landscaping involving an AI which learns the decision making patterns of experts in a flexible and efficient manner to classify a vast amount of data. We shaped the technology so that the AI can accurately pick up subtle signals made by a decision maker. Even though the AI has access to only a small amount of training data*, it can still perform as proficiently as an expert. After having created the technology of Landscaping, a machine learning and natural language processing solution that can evaluate data on a par with a human expert, we shifted our focus to figuring out how the software can use such knowledge more efficiently.

  • *training data: a data set used by the AI to teach itself how to classify data.

How does UBIC use expert knowledge?

In order to utilize expert knowledge more effectively, it is not only necessary for the software to have the ability to acquire knowledge instantaneously and flexibly regardless of circumstance, it should also have the capacity and endurance to build up a knowledge base for the long run. This is because users can store their valuable know-how of information analytics into such a structure. UBIC's knowledge base takes in more than 1,500 past investigation cases. Word combinations, typical keywords related to cross border litigation or anti-trust investigations, and typical behaviors, are all found in data readily available for users.

We propose several variations for creating knowledge databases. For example, the knowledge base can be created by inputs from a talented expert who is considered an authority in his or her profession. Also, there is another approach which is designed to generate a repository of collective opinions and insights gathered from multiple experts. If you happen to take a look at the data inside a knowledge base, you will find a fusion of concepts which are quite difficult to describe by words. We store this as tacit knowledge. Tacit knowledge proves to be useful when finding the wanted data by means of an on-site data analysis.

Application of tacit knowledge typically works well when carrying out on-site analysis because it can fulfill the requirement of finding the target. Consequences related to the discovery actually does not matter much in this situation.

However, when we analyze information to help improve various business activities, the method of categorizing and explaining data according to each purpose proves to be useful. As our goal, we believe that not only should we categorize data, we should aim to motivate our clients so that they could take further actions to build a brighter future. At UBIC, the concept of Behavior Informatics is applied to the system’s framework of data categorization. Moreover, we integrated into our models areas which prove reliable for enhancing our understanding of the data―namely social science, criminology, and psychology. For example, internal corporate fraud usually does not happen out of the blue. Before it actually occurs, it usually goes through a predictable sequence of events. After conducting a series of analyses both quantitative and qualitative on this event sequence, we uncovered a recurrent structure and crystalized it as a model. We then linked relevant data to the model. In this way, Behavior Informatics Laboratories evolves by combining methods from various fields in its research and development.

So you ask, what is KIBIT?


KIBIT is a manifestation of Behavior Informatics.

KIBIT is an AI-based search engine that can make predictions and informed decisions while classifying data.

KIBI is driven by algorithms such as Landscaping, machine learning, and NLP, and also has the ability to use its knowledge bases in order to respond adequately to various circumstances.

KIBIT features a versatility that can support almost any kind of business scenarios and circumstances.

For example, even though legal affairs or medical clinics are places where only qualified professionals can practice, KIBIT can still step in to help Attorneys and Doctors analyze their data with ease so that they can manage a successful eDiscovery or deliver illness diagnoses.

On the business front, KIBIT can help by providing updates to managers who are trying to close deals or lead projects successfully. Mangers can take advantage by having KIBIT help monitor their project progress or sales activities.

Another area elegantly navigated by KIBIT is the digital curation services. You will be surprised by the wonderful recommendations and ideas KIBIT will offer you by finding information truly reflecting your likes and dislikes. KIBIT is always keen to learn more about you, too.

We named our search engine KIBIT.It derives from the Japanese word "Kibi (subtlety)" and "bit" the smallest unit of information in computing.

Becoming an "AI that can understand the subtle behavior of human beings", we hope KIBIT matures by fulfilling the role of supporting various people and asserting a powerful influence across various business fields.

What is Behavior Informatics Laboratories?

Our lab carries out research and development based on the concept of Behavior Informatics. We conduct research―in the areas of natural language processing, machine learning, data mining, behavioral science models, performance optimization, and building knowledge bases―while providing support to our clients in their business activities by delivering them our research results and findings via software applications and cloud infrastructure.

We give special attention to how our innovations and technology could benefit our users who are also involved in information analysis. For this reason, the typical time span from basic research to product development is short, and our agile workforce offers our clients adequate service swiftly.

With an aim to help society become safer and better,
we take part in creating the future with Behavior Informatics.