FRONTEO Co., Ltd. (Headquarters: Minato-ku, Tokyo, President: Masahiro Morimoto, hereinafter referred to as FRONTEO) has a self-developed AI review tool "KIBIT Automator" that allows humans to see colors at a glance. We are pleased to inform you that we have implemented a new algorithm (patent pending) inspired by the excellent intuitive judgment ability when recognizing.

 

　In recent years, with the rapid development of the digital society, the use of IT has become commonplace, and the amount of electronic data managed by companies is steadily increasing.In e-discovery (electronic discovery) in international litigation and digital forensics, reviewing documents related to evidence from such a huge amount of data is a difficult task that requires both speed and quality in a limited number of days. However, quality variations due to the skills of reviewers, increased work time and huge costs due to the amount of data are major issues.

 

　KIBIT Automator is an AI tool that contributes to significantly improving the efficiency and quality of review work.So far, this product has an algorithm that has a function to calculate the relevance of the words in the document to the evidence as "weight" (degree of relevance of the word to the evidence) and a relative accuracy of document discovery related to the evidence. It was equipped with two types of algorithms, one with high accuracy and the ability to accurately predict discovery accuracy before review.These functions have brought about effects such as standardizing quality, improving efficiency, and greatly reducing work time and costs for the enormous review work of companies and lawyers involved in litigation.

 

　The new algorithm implemented this time combines the strengths of the two conventional algorithms, and has even higher accuracy and a high degree of explainability for the judgment basis.When reviewing documents, humans have the ability to quickly and accurately determine whether evidence is relevant or not.Inspired by the fact that this is similar to the ``color bus effect,'' which is a visual intuition that quickly and unconsciously recognizes only what it wants to detect, he said, ``Humans perceive words in a document as colors, and the nuances of the entire document. It was developed as an algorithm that can make judgments that are close to the intuitive judgments of humans, based on the idea of ​​a cognitive model in which the number and combination of word occurrences are captured like colors.

 

　With the new algorithm installed, KIBIT Automator can now indicate the degree of "negative" relevance in addition to the conventional "positive" as a word weight evaluation.In addition to the score of the entire document (degree of relevance to the evidence), it also has a function to calculate the score distribution within the document and identify and highlight areas with high scores.These functions contribute to increasing the explainability of why AI judged a document to have a high score, which is generally considered difficult to identify.In addition, in an in-house verification using the regular audit tool "KIBIT Eye", which already has this algorithm installed, we compared the extraction rate (discovery accuracy) of the data to be audited at a viewing rate of 20%. At the time, it was confirmed that the analysis accuracy was greatly improved to 82.5% with KIBIT Eye, compared to 99.5% with the conventional system.In addition, since the new algorithm is highly versatile, we believe that it will be possible to deploy it in various FRONTEO products in the future and improve their performance and functions.

“KIBIT Automator” is an AI tool that supports document review in electronic discovery (e-discovery), which is required in the court proceedings of US civil lawsuits. We use AI to examine and analyze large volumes of e-mails and electronic files, which are evidence materials, and contribute to improving work efficiency, reducing the burden on workers, and reducing costs.

 

■ About FRONTEO　URL: https://www.fronteo.com/