“We are providing technologies for taking exams on the Internet for companies which allows us to test reliability and validity in a short time. We offer programs such as ‘English Navi’ for Eiken Society, the Japanese English Test Association, which is our pillar of revenue from technology licensing. We would like to advance the development of artificial intelligence (AI) technology to capture the demand of the education industry that is lagging behind in IT. We are developing our own AI to incorporate in our program, and we can provide flexible systems according to customer’s needs. Taking advantage of the character recognition technology cultivated through the development of AI technology and automatic scoring technology for handwritten tests, we will capture the demand for automation of increasing administrative work in the financial industry, insurance industry and healthcare industry,” comments Junichi Takamura, founder and CEO of EduLab (TSE: 4427).
EduLab is Japan’s leading EduTech innovator in e-testing and e-learning powered by its adaptive test technology and artificial intelligence (AI) OCR handwritten character recognition technology “DEEP READ” that enables it to achieve high profitability with operating margin of 24.3%, ROE (= EBIT/Equity) 38.1% and ROA of 12.4%. DEEP READ is also incorporated as a solution in Box Inc. (NYSE: BOX) since Aug 2018 to expand into other fields/markets. EduLab provides test development, implementation, analysis, educational services of English and other competence tests through its focus on the R&D of ability measurement technology and educational tools in the education field for customers such as public examination organizations (Eiken Society/ Japan English Test Association; schools; MEXT/ Ministry of Education, Culture, Sports, Science and Technology; local governments; general corporations). EduLab, the parent company of JIEM (Japan Institute for Educational Measurement Research), is the Japanese equivalent of the Educational Testing Service (ETS) in the U.S. which is the world’s largest educational testing and assessment organization. EduLab is also the creator of CASEC (Computerized Assessment System for English Communication) which is Japan’s first adaptive English test developed since 2001 and the number of examinees is over 2 million people (total number of examinees in tests contracted to JIEM is around 17 million people), and is also entrusted with the test scoring work for governments and educational institutions. EduLab operates two business segments:
(A) e-Testing/e-Learning (74% of sales): EduLab provides educational services for educational institutions such as universities, private enterprises and individuals with the Eiken Society (a public interest foundation established in 1963 for the purpose of dissemination and improvement of English) as the main customer. Direct sales to the Eiken Society occupy 39.2% of the company’s sales, including the educational digital solution. About 60% of sales depends on the Eiken Society. Main digital services are:
(1) English Navi & Study Gear (49.3% of FY2018/09 sales, +25.1% yoy) provided by the Eiken Society and Educational Digital Solutions to support English learners including examinees for the Eiken exam in which more than 22,000 schools and cram schools participate and more than 400,000 people are continuously learning;
(2) Providing test system (29.1%, +48.5% yoy) such as the Eiken-kai 4–5 class speaking system;
(3) CASEC (11.7%, +3.7% yoy), an IRT (Item Response Theory) computer-aided test in joint development with the Eiken Society which can accurately measure the test examinee’s ability regardless of the difficulty of the test by automatically changing and analyzing test questions according to the individual’s abilities and is the world’s first test using adaptive technology;
(4) Eiken Jr. (7.1%, +13.1% yoy) which is an English education and testing program with schools and cram schools commissioned by the Eiken Society to run and operate; and
(5) TEAP CBT (2.8%, -29.6% yoy) (Test of English for Academic Purposes — Computer Based Testing) which is jointly developed and operated with Eiken Society and Sophia University, and mainly intended for recruitment at educational institutions such as universities that conduct university entrance examinations for high school third graders e.g. 27 schools in the private Sophia University.
(B) Test operation & Contract (26% of sales): EduLab is entrusted with the creation, scoring, aggregation and system construction, management and operation of test questions from national and local governments that implement academic ability research projects, as well as public institutions, universities and other educational institutions, such as the Eiken Society, achievement survey (PISA), international adult competence survey (PIAAC), international mathematics and science education trend survey (TIMSS), Saitama prefecture academic ability survey utilizing IRT, etc. Since Jan 2018, the AI OCR automatic scoring technology was utilized for “English 4Skills” learning service developed and operated jointly with NTT docomo.
There is an enhanced English learning demand for qualification and examination test in a paradigm shift in the education industry with the structural change of the external educational environment that include a new university admission common test from the fiscal year 2020 to include the qualification and examination test of the four English skills, adding ‘talk’ and ‘write’ to the two skills of ‘read’ and ‘listen’; introduction of digital textbooks and CBT of various tests at primary, middle and high school; conversion of textbook into English curriculum in the fifth and sixth grade of elementary school and English is compulsory in the third and fourth grade.
EduLab’s business model until 2015 was a contract type that does not incur business risk after delivery of digital services, receiving contracted development fee, maintenance and operation fee. It was a relatively simple profit model that sales would increase if the number of entrusted projects increased. Since around 2015, EduLab has been transforming to a profit-sharing business model, partnering up and taking business risk to grow with its customers (mainly the Eiken Society), receiving performance-based license fees linked with the number of candidates, the number of organizations, the number of months of usage. Given that many domestic educational institutions have their fiscal year end as of March, there is seasonality in EduLab’s business in which sales tend to concentrate towards the second quarter and fourth quarter.
EduLab achieved a 15% yoy increase in sales and 63.9% yoy rise in operating profit in FY2018/09 with improving operating margin from 17% to 24.3% and ROE (= EBIT/ Equity) of 38.1% and ROA of 12.4%, propelling a 11.6% increase in market value since its listing on 21 Dec 2018 at the IPO price of 3,270 yen per share to US$273m. Balance sheet is moderately geared with net debt of 1.413bn yen (gross cash 3.125bn yen, gross debt 4.538bn yen), or 4.5% of market value. Management expects FY2019/09 sales to rise 71.6% yoy to 6.81bn yen and operating profit to increase 46% to 1.4bn yen.
NTT DoCoMo and Obunsha, a major educational publishing company founded in 1931 with its English textbook for college entrance examination achieving #1 in bookstore sales, are strategic shareholders with 5.06% and 6.34% respectively. Together with NTT docomo, EduLab provides “English 4Skills” for college entrance examination and AI automatic scoring technology. EduLab also provide subjects other than English in a joint project with Obunsha in offering free animation-based multi-disciplinary learning application for elementary and junior high school students under the ‘Study Gear’ brand.
EduLab also provides achievement tests for elementary and junior high school students in self-study classroom projects in China and India. In April 2018, EduLab established a venture investment base in Boston to invest in EdTech venture companies. EduLab has adopted a stock option scheme to incentivize and align the management and staff. The number of potential shares by stock acquisition rights is 1,024,400 shares, which is equivalent to 13.58% of the total number of issued shares of 7,542,600 shares. EduLab was founded in 2000 by Junichi Takamura and Hiroshi Matsuda.
DEEP READ competes with Japan’s Cinnamon which claims around US$3m in revenue and over 200 clients. Founded in Oct 2016 by Miku Hirano, Cinnamon’s core SaaS cloud product is Flax Scanner, an AI auto-documentation tool launched in April 2017 that understand context and reformat the documents from unstructured business documents and email for both handwriting and text data. Cinnamon claims high recognition accuracy for hand-written reading of 95–98% for real business data. Cinnamon has raised over US$24m. DEEP READ differentiates from most AI OCR companies who do not have handwritten character recognition technology, such as NZ-based Xtracta (~US$4m revenue), Singapore-based TAIGER (~US$6m revenue), Germany’s ITyX Solutions (~$10m revenue). AI OCR companies with handwritten character recognition tech include Japan’s Cogent Labs and San Jose-based Infrrd (~US$5m revenue).
CEO Takamura comments thoughtfully that there is a paradigm shift and structural change in the educational environment in Japan that creates an enhanced English learning demand for qualification and testing: “As globalization accelerates and more practical communication skills in English is required, the course of English study has been revised for the first time in about 10 years recently, and a new university admission common test to include the qualification and examination test of the four English skills, adding ‘talk’ and ‘write’ to the two skills of ‘read’ and ‘listen’ so far. Therefore, the teacher needs guidance not only to teach ‘read’ and ‘listen’, but also to give power to ‘talk’ and ‘write’ more than ever. In addition, the government is promoting the use of ICT in school education, and it is a policy to organize measures for installing one information terminal per one school. It is expected that this will require digital teaching materials to be enhanced using information terminals. English education reforms are advancing at a rapid pace and the number of candidates for the Eiken practical English proficiency test has steadily increased. There is an enhanced English learning demand for qualification and examination test in a paradigm shift in the education industry with the structural change of the external educational environment:
(1) Under the development of the ICT environment at schools conducted under the 5 Year Plan 2018–2022 for the ICT conversion of education, it is expected that ICT will be actively utilized such as the introduction of digital textbooks, CBT of various tests at primary, middle and high school, including the English speaking CBT National Academic Achievement Survey conducted nationwide in 2019.
(2) From the fiscal year 2020, it is expected that the ‘university admission common test’ will change greatly including the embodiment of the evaluation of the four skills of English (reading, listening, talking, writing). It is difficult to concurrently conduct the tests on the same day in large group at the exam center. Therefore, it is suggested to utilize private qualification and certification test which is already evaluating the English 4 skills, such as Cambridge English tests, TOEIC, TOEFL. By 2024, there will be a complete transition to English private examinations.
(3) Due to the revision of the guidelines for teaching guidance, textbook will be converted into English curriculum in the fifth and sixth grade of elementary school and English is compulsory in the third and fourth grade.
We believe that future market movements will be a major business opportunity for EduLab who will become indispensable in the evaluation and measurement of learning progress in the knowledge-based society.”
When asked about how he came about to start EduLab, the challenges he faced and overcame and the breakthroughs during his early entrepreneurial period, CEO Takamura shared briefly and reflectively the story of his early entrepreneurship period: “After completing an MBA at the University of Hartford, USA, in 1990 and an MS at the MIT Sloan School of Management in 1993, I was engaged in the high-speed scanner business at Scan-Optics LLC. I joined the Hughes Electronics Corporation in the US in 1993 and worked as a business development manager in a satellite-related business focusing on the establishment of DIRECTV Japan. After leaving the company, I joined an executive search firm Jomon Associates, Inc. and worked there for about 3 years, recruiting executives for both Japanese and foreign businesses. I started an independent business consulting practice in 1999 and joined Obunsha Group with a mission to develop and promote new businesses.”
“I was involved in the development of new business by the Eiken Society (Japan English Study Association) in 2000 and planned the establishment of the company Evidas that provides educational services via the Internet. Our co-founder and current COO Mr. Hirofumi Matsuda participated in Evidas in April 2000 to support the opening of the English learning portal site Evidas. Matsuda had completed his Master’s Program in Keio University’s Graduate School of Media and Governance and worked as a graduate school researcher at Keio university and was the leader in the IT department at Obunsha Enterprise.”
“EduLab was founded in March 2003 by the stock transfer of Japan Institute for Educational Measurement (JIEM) which was established in October 2001 with the R&D members of the Eiken Society, and Obunsha supported the establishment of it. In March 2002, Obunsha acquired all shares of Evidas, and Evidas acquired all shares of the Institute for Educational Research and Measurement. In May 2002 Evidas was merged with Education Measurement Research Laboratory with the aim of integrating Internet-related technology and test-related technology. EduLab has since advanced its business in the field such as CBT (Computer Based Testing) service and educational testing solution business.”
“Mr. Norio Hayashi, our third largest individual manager-shareholder, had joined the Eiken Foundation of Japan in 1982. He was engaged in research, education, and the application of test theory which is the measurement technique in the field of education. He also established the practical use of a latest testing theory. He led the development of the next generation English test, CASEC (Computerized Assessment System for English Communication), which was released in 2001. He has been a director of the foundation since 2003. He joined JIEM in 2002. He is presently serving as a director of the Japan Association for Research on Testing.”
On the strengths and strategy of EduLab, CEO Takamura comments: “EduLab’s features and strengths include:
(1) state-of-the-art technological capabilities in the educational IT field. We have differentiated from other companies by taking advantage of accurate ability measurement technology. Examples of state-of-the-art technological capabilities include IRT (item response theory) and CAT (computer adaptive) technology that lead to the development of CASEC. In addition, we are developing a highly accurate handwritten character recognition technology ‘DEEP READ’ using AI technology, which we are applying to the descriptive answer scoring process and to other areas.”
(2) alliances with strategic business partners. We have a long-term relationship with the Eiken Association, which is a representative corporation that works to train English proficiency, as a strategic business partner,. In addition, we have long-term partnership with Obunsha in the educational contents related business, and NTT docomo in the ICT education related business.
(3) intellectual capital such as knowledge and information gathering abilities in overseas EdTech markets.”
“EduLab’s basic strategy of expanding the scale of its business and ensuring stable earnings is: (1) Product development to capture business opportunities by improving the skills of English exam; (2) Expansion of learning tools for preparation of test; (3) Implement various tests, expand the scope of infrastructure, cultivate customers, (4) Investment of EdTech to become the core of the next generation education solution.”
When asked to elaborate on how EduLab’s AI technology DEEP READ came to be developed, its potential applications and expansion plan, CEO Takamura shares: “Various academic achievement surveys are shifting to a method of comprehensively evaluating ‘knowledge / skill’ and ‘critical thinking ability, judgment ability, expression power’ from the conventional method that mainly focuses on just ‘knowledge and skills’. But the scoring costs accompanying this trend tend to increase. The reason behind the development of DEEP READ was the enormous amount of time and cost savings associated with grading of the answers that had been a long-standing problem.”
“And the need for automatic scoring is increased by the rise in the number of descriptive papers from the 2020 college entrance examination exam. EduLab has developed and launched on 18 Jan 2018 a highly accurate handwritten character recognition technology called ‘DEEP READ’ using AI technology based on deep learning from the viewpoint of improving grading efficiency of descriptive expression solutions in large-scale academic ability surveys and tests, shortening scoring man-hours and the scoring lead-time. EduLab intends to differentiate from our competitors by applying the ‘DEEP READ’ to realize the innovation of the grading process of the written formula solution.”
“Many companies have already worked on character recognition itself by utilizing OCR technology. With traditional OCR technology, we are making databases of various fonts. However, it may not be compatible because there is low recognition accuracy of uneven handwriting. Hence, relying on human for data input work was common. Many enterprises are handling BPO outsourcing overseas for inputting handwritten characters, which is incurring enormous time and cost. There are many companies that manage vast amounts of data on paper, such as the customer data and various application forms in insurance companies, financial firms and medical certificates in hospitals.”
“In order to solve these problems and to increasing the efficiency of our educational content and scoring work, we have started to research and develop AI character recognition for about 2 years. DEEP READ is a character recognition technology that utilizes AI (artificial intelligence) based on deep learning, which is distinct from conventional OCR technology. When we were developing DEEP READ, we initially tried recognition by each character which is what our competitors are currently doing. However, in the case of handwriting, handwritten characters are connected to one another and the boundary of the characters to each other was unclear. Also, if there is a ruled line, it will be difficult to separate letters if it overlays a border line. Furthermore, there are countless numbers of characters that are difficult to distinguish.”
“In addition, because the Chinese characters is constituted by a combination of characters, there was a problem if it is taken one character at a time. Our AI make it recognize handwritten characters including not only each character but also the context of the text before and after, as well as by analogy from the flow of sentences and the relationship of letters. We succeeded in raising the accuracy of handwritten character recognition by letting the AI learn the context of sentences with the same flow as human beings are reading them. This point also gradually improved with the approach by AI, and now it is possible to read multiple characters with very high precision.”
“Currently, the character recognition accuracy of DEEP READ is 94.5% (without borderlines which is closer to actual answer sheets) compared to the average of 83%, and 98.9% accuracy when reading specific-format handwritten forms. Up to 200 multiple characters can be recognized at once. There are companies that claim higher accuracy among competitors, but handwritten character verification data samples are few and credibility is low in many cases. The accuracy of DEEP READ is based upon data verified using 35,000 test data of all industries such as finance, education, and government. Also, we verify using the actual customers of our company about the answers of the exam, and we are drawing out overwhelmingly high figures. DEEP READ also develops English version. Accuracy is higher than Japanese because of the alphabet only, achieving an average accuracy of 97%.”
“As an image of DEEP READ to be used at the educational site, the handwritten answers are converted into texts by AI and read into the scoring system, and the answers are classified in advance according to contents including or not including specific keywords. Regardless of whether the answer is correct or incorrect, we will be able to drastically reduce the time of the grader who had to confirm all answers.”
“From Aug 2018, we collaborated with Box to utilize DEEP READ handwritten recognition service utilizing AI to automate the digitization of handwritten characters and make search in the cloud possible, so that it can be used for more efficient document management. The content management service Box is a platform that allows you to securely manage and share files on the cloud, as well as create and edit documents by multiple users. Currently, digitization of the image data of the imported document is by conventional OCR (optical character recognition), but there was a problem in that the recognition accuracy of handwritten characters was low.”
“At the BoxWorks2018 event held in San Francisco, EduLab exhibited and demonstrated DEEP READ. Many government officials who stored many old documents and companies that have problems in digitizing handwritten application documents visited and we received very favorable reviews. EduLab has transferred the development base of DEEP READ to Boston and will further deepen the R&D of the AI project. DEEP READ enables digitization of handwritten characters with overwhelming superiority in accuracy of over 94.5%, which is the top level in the industry and 83% on average for others, and is widely used by companies such as the financial industry, medical institutions and school corporations.”
“In addition, we believe that this character recognition technology is applicable not only to educational IT field but also in a wide range of industries processing enormous paper-based data, forms and applications such as banks, insurance companies, hospitals. To this end, EduLab established DoubleYard, Inc. in Boston in April 2018. Regarding business development, we have started activities not only in the Japanese market but also in other markets including China and India market, and as a result, we have been engaged in large foreign financial institutions, major newspaper companies, major BPO companies, and government. We have also entered into collaborative projects with relevant organizations, universities, etc.”
“As preparation for large-scale order client acceptance in the future, we are now preparing the API environment with the goal of starting operation soon. As a mode of providing this service, we plan to use two types: cloud software-as-a-service pay-as-you-go billing model and on-premises system solution model depending on the server usage. The basic service of the cloud type is from 200,000 yen a month, and it can add functions according to the user’s request. We aim to introduce 30 to 50 companies a year. Because the technique and service development method utilizing deep learning is relatively easy to apply to other fields, EduLab has developed capabilities that utilize the AI cultivated through the development of handwritten character recognition technology to develop in other fields and enhance the competitiveness of the products and services of the entire Group.”
“We are considering natural language processing and personal adaptive learning management system engine as the areas where EduLab is planning to utilize AI. Through these developments, we believe we can improve the competitiveness of our products and services and expand profit margins in all segments of the Group. Regarding the areas of application of the AI handwritten character recognition technology developed by the Group, we intend to actively pursue collaboration with partners of other industries and to expand applications to other industries.”
“On 22 Jan 2019, we collaborated with TIS Group’s (TSE: 3626) Agrex Corp, a major BPO information service company, to utilize our AI OCR DEEP READ from April 2019. By combining our AI technology with their accumulated data processing technology know-how in BPO business for over 50 years, we will provide accurate and efficient quality data entry service and create new AI x RPA solutions. Financial companies, which account for about 60% of the customers of Agrex, are moving towards automation by AI x RPA, while dealing with sensitive personal information such as bank information and credit card information. We will utilize DEEP READ in on-premise environment so that we can handle the outsourcing. AI-OCR engine-equipped server is installed in the Agrex data center and data processing using DEEP READ is realized without going through the network. In addition, since we do not build the environment on the customer side, it is possible to reduce introduction and maintenance cost.”
“EduLab was recognized for the first time in Japan in July 2017 as an artificial intelligence company in the education field for our handwritten character recognition based on deep learning technology in the NVIDIA Inception Program’ which aims to train start-up companies trying to revolutionize the industry with advances in AI and data science.”
Curious about EduLab’s different services, we asked CEO Takamura to elaborate more and he shares patiently: “English Navigation · Study Gear is a practical English proficiency test to support English learners including those who take exams. Through the measurement → diagnosis → learning cycle, it provides the best teaching materials for each individual, provides the latest information useful for English learning, supports educators and learners through data analysis, and also serves as a platform for online application for Eiken exam.”
“In addition, we offer online study service ‘Study Gear for EIKEN’, the Eiken official online learning app for English learning. Study Gear for Eiken has the concept of ‘want to start, want to continue’, and candidates aiming for Level 2 to 5 of Eiken are able to learn all the necessary knowledge on each level online in this fun learning tool. Since the launch of the service in 2013, we received high praise from users from not only individuals, but also many groups such as schools and private schools, and the number of users is over 22,000. There is also data that the pass rate will increase by approximately 20%.”
“While mobile study apps have become popular in Japan, EduLab noticed that many of them are not only specifically aimed at older students studying for university entrance exams, but they also charge for their services. Seeing a need to help younger learners, EduLab has released Study Gear, a free multi-disciplinary app service offering compact lessons for primary and younger junior high school students since 2003. Subjects covered are Japanese, mathematics, society, science and English, all of which can be learned via short animations. The one to 2½ minute videos are made by the One Media video production company, while publishing house Obunsha provide the topics and educational information. As a bonus to using the app, children can also rack up NTT docomo loyalty points for their parents each time they complete lessons. There are currently over 3,000 video lessons available on both iOS and Android versions of Study Gear. We plan to earn early monetization by acquiring personal and corporate users and advertisers of the platform through appropriate marketing strategies.”
“From 2 April 2018, together with NTT docomo, we offered the learning app service ‘English 4Skills’ using smartphones, tablets, PCs and a test that can easily measure the skill of each student while students advance their learning at their own pace. NTT docomo has begun rolling out a nationwide 5G network across the country and has chosen EduLab as a partner, giving us the opportunity to test the delivery of educational content and services on this new high-speed network.”
“Eiken Jr. is an English education and testing program conducted by the Eiken Society that covers schools and cram schools for elementary school children, and EduLab is entrusted with its management. Eiken Jr. is aiming at studying and researching the English proficiency of the students, with the goal of students becoming familiar with English and gaining understanding of foreign cultures. The listening test is a ‘training type game sense’ for children.”
“We provide Eiken-kai 4–5 Speaking System, a service that provides English tests to measure the speaking ability of the 4 and 5 class applicants. Individual can make payment at convenience stores, via Eiken organizations (mainly schools), and payment by credit card.”
“CASEC (Computerized Assessment System for English Communication) is the world’s first test using the IRT (item response theory) and CAT (computer adaptive) technology developed jointly by the Eiken Society and EduLab. IRT accurately measures the ability of a test taker without being influenced by the difficulty of a uniform test item like a conventional test, and automatically analyzes the answers to the test items of the test examinee using the adaptive method.”
“CASEC is Japan’s first adaptive English test developed since 2001 with 4 million answers and the number of examinees is over 2 million people and the number of candidates increases year by year. In the modern Japanese society in which further expansion of English ability is an extremely important task, we take advantage of the academically supported measurement theory such as IRT (item response theory) and computer adaptive test to differentiate CASEC as a next generation English test that can also be taken 24 hours on the internet. The adaptive test which automatically changes and analyzes test questions according to the individual’s ability makes it possible to compare the test results of examinees who took different tests based on the examinees’ answers and to qualify an optimal level of problem according to the examinee’s ability for each question, accurately measuring the examinee’s ability in a short time. Accuracy and speed in adaptive testing are the unique value proposition of CASEC.”
“In addition to individual examination, we have proven to introduce to more than 1,000 companies and more than 200 universities, and in domestic English tests we are proud of the number of exam candidates. The TOEIC and estimated Eiken standard score are also shown in the test results which are displayed immediately. Introduction is being introduced as a means to accurately and easily measure English proficiency at companies, universities and educational institutions. Major companies such as major travel agency HIS (TSE: 9603) and Rakuten also introduce CASEC for employee English education and to create an internal corporate culture to promote English learning, maintaining a high examination rate at all times. CASEC is also used to select overseas assignees.”
“Since offering CASEC as Japan’s first IRT adaptive computer test, we have innovated further. in November 2005 we introduced CASEC-G which automatically scores English sentences; in April 2006 English grammar learning teaching material CASEC — GTS; in November 2010, we developed and started operation in a new learning system CASEC — WT to improve English writing power. In 2018, we began offering CASEC SPEAKING, an English speaking ability measurement service for beginners developed with AEON Research for schools, English conversation schools, enterprises, and we are introducing in AEON’s English conversation classrooms in May 2018. AEON has 250 schools in major cities throughout the country, and the environment and curriculum make it easy for students and society people to communicate and learn.”
“The ability to ‘speak’ is difficult to measure. There are TOEIC and GTEC in the private speaking test in which the examination dates and times are specified, but there are not many tests that specialize for English beginners to measure the speaking ability at this level. There are also cases where the examination fee is expensive. In comparison, the exam fee for CASEC Speaking is 4,500 yen. In the future we plan to allow individuals to take the CASEC Speaking test at any time from their home PC. In the score report, you can learn outcome such as pronunciation, fluency, vocabulary power, grammar power.”
“CASEC SPEAKING aims to connect to the next learning while maintaining motivation even for beginners who speak, such as setting the difficulty level. This will create a cycle of continuous learning and measurement, aiming to improve the speaking ability of English learners.”
“TEAP CBT (Test of English for Academic Purposes) is an English proficiency test jointly developed by Sophia University and the Eiken Society and EduLab. TEAP CBT is intended for recruitment at educational institutions such as universities that mainly conduct college entrance examinations for high school third graders. The contents of the questions are created taking into consideration scenes encountered in university education, including study abroad. TEAP CBT is a test to measure four skills of English ability: reading, writing, listening, speaking.”
“Our Test Operation & Contract business provides all the functions required for the execution and operation of the test, including the creation of test questions, system construction, scoring & analysis, management, operation, etc. Customers are public institutions such as Eiken Society, university educational institution, and local public entity that conduct the academic ability research project. In addition, we undertake content development and system development in test development (e.g. TEAP CBT, Eiken Jr.) by the Eiken Society etc. We offer test analysis and consulting services for educational institutions and private companies.”
“For instance, Saitama prefecture conducted an academic achievement survey in Jul 2018, which is a groundbreaking study that measures the ‘growth in academic ability’ of individual students utilizing IRT. Along with the nationwide academic achievement survey, each prefecture often carries out academic research on its own. Data of about 300,000 students are collected and analyzed.”
“It is extremely rare worldwide to implement such a survey on a large scale at the municipality level. If you want to measure the growth of the same child, it is not enough to do the usual standardized examination such as TOEFL. For example, let’s say that a child achieved a high score. However, there is a possibility that the test may be simpler. In order to regard the result of the test as a proof of growth, it is necessary to consider the difficulty difference of each test. If you only see if it is higher or lower than the average in that same group, you can see by the deviation value or the average point. However, whether or not a person’s ability has increased has not been known unless the ‘measure’ is the same.”
“It is important to be able to accurately determine the difficulty level of the problem. Every time a fourth grade student progresses to fifth grade, follow-up survey is carried out. By doing this you will know how much power the child has put up. This is a big difference with the conventional academic ability survey. That work is the statistical method IRT. Understanding and sharing to teachers including the principal was not sufficient. The school organization is accustomed to the idea of the average point for a very long period of time and it takes time to release them from the spell that ‘My school’s average score in the academic achievement survey is plus 3 points from the prefectural average’. Fukushima Prefecture also participates from next fiscal year.”
“After analysis based on survey data, we are also examining the mechanism by which children grow. As a result, positive attitudes are spreading, such as sharing tips for extending the power of children among teachers. if you can improve elements and unrecognized abilities such as self-discipline, diligence, feelings to work hard even if you are not good at it, your academic ability will grow.”
“The Saitama prefecture groundbreaking survey also drawing attention from the world. The OECD, which 35 countries around the world are members, wants to know the conditions of teachers who will increase the students’ unrecognized abilities. However, the OECD’s Position Achievement Survey (PISA) is not a continuous survey for the same child. Therefore, it is pointed out that there is a limit in grasping the condition of a good teacher even when tied to the international faculty guidance environmental survey (TALIS). So OECD began to pay attention to Saitama prefecture academic curriculum. OECD Education and Skill Director Andreas Schleicher, also known as ‘Mr. PISA’ for devising the PISA survey, visited Saitama Prefecture pan in July 2017, and he insisted on the necessity of collaboration. There is the possibility that the new academic ability research created by Saitama Prefecture will become an influence on the educational policy of the world in the future.”
“In the future, AI gathers the results of answers and AI presents the best exercise questions for individual children through analysis to support teachers. In the Saitama prefecture survey, it is shown that non-cognitive ability is cultivated in the relationship between children and teachers. Regardless of how much AI develops, the essence of learning and the necessity of teachers that people will learn together under the teacher will remain unchanged. The new development in big data analysis and the latest technology that objectively grasps the academic ability and non-cognitive abilities of children puts forward a stone to pave the way of what the future education will be.”
CEO Takamura also commented on EduLab’s overseas business expansion in China and India: “EduLab started a cram school project using ‘adaptive learning’ in China with a partner company in 2014 and started service in the summer of 2017. After the preparation period of the system and contents for about two years, we started the ‘reverse learning’ course in two subjects of mathematics and national language. It is a new blended learning concept in which the students prepare for new learning contents usually by watching short anime video lessons at home, and teachers do not give lectures in the classroom and provide guidance tailored to each student about the subject, and the students work in cooperation with other students.”
“With adaptive learning using online tests, it becomes possible to accurately identify where each student stumbles in the contents of learning and where understanding is insufficient. You can proceed to the next step while confirming the degree of learning progress and comprehension degree, and you can accumulate data of each student individually, so you can improve the quality of service individually. We operate a directly operated school in Wuxi, Jiangsu Province and Changsha in Hunan Province, and promoting business alliances in other regions areas such as concluding regional agency contracts.”
“As of December 2018, the number of deployed cities throughout China is 47 cities, the number of classrooms we manage is 117, and the number of paid students registered is about 6,000. Teaching materials are provided both online and paper media. Business revenue has directly managed store operating revenue, member school registration fee and royalty revenue by providing teaching materials to member schools and system. The reason why the adaptive learning study room is beginning to be accepted into the Chinese market is that we are developing original teaching materials for China from scratch. The results of the study room are visibly reflected in the results at school and the system grasps each student’s abilities to maximize student learning effect while reducing teachers’ load.”
“The mainstream traditional cram school is one where one teacher taught dozens of students. Therefore, the level and popularity of the teacher influenced the management of the cram school, and securing and maintaining the charismatic teacher puts great pressure on the management of the cram school. Since the study room has a greater emphasis on the design of the teaching materials and the system itself, the role of the teacher is mainly the ‘coaching’ that was appropriate to the students as compared to ‘teaching’ to the group. As teachers play a major role in student’s learning progress management and motivation maintenance, it became possible to reduce work while offering high quality guidance to students.”
“As future management tasks, we will promote further expansion of business including junior high school student business development, expansion of Shanghai teaching material development base and Wuxi system development base, strengthening internal management system, introduce adaptive engine with AI function to improve the accuracy of analysis of personal learning history data by installing the personal adaptive learning management system in the existing system. We aim to achieve 100,000 students.”
“In our India Kyoshi project, EduLab provide achievement tests for elementary, middle and high school students. We concluded an exclusive sales partnership agreement with Smartclass Educational Services, which has a direct marketing network nationwide to elementary, junior high and high school. We have sold approximately 12,000 tests. Future management tasks include strengthening the test development base in Pune in India, converting Bangalore logistics base nationwide to a logistics center, deploying our proprietary item response theory (IRT) CBT test.”
CEO Takamura went on to share about EduLab’s EduTech investments in CodeMonkey, SpeakingPal, Authess, Listenwise, Academic Merit (Finetune) and ELSA: “Over the past few years, EduLab has invested as a limited partner in three U.S.-based education technology venture funds: LearnLaunch, Fresco Capital and GSV Acceleration. With the experience, EduLab has established EduLab Capital Partners based in Boston in March 2018 to be a general partner in creating our own edtech fund to focus on investments at the seed and Series A stages, with a specific interest at high-growth edtech and artificial intelligence-related startups. We have a small AI team in our Boston office that may work with some of the portfolio companies. We established EduLab Capital Management Company, LLC in Boston as a subsidiary of Edutech Lab, Inc. in America with the aim of accelerating investment in EdTech venture companies. In the past few years, we have invested in 4 companies in the US, 2 in Southeast Asia/China, 2 in Israel. These include Listenwise, a listening comprehension service that uses public radio programs as source materials; Authess, a digital, performance-based assessment tool; Academic Merit; ELSA; CodeMonkey and SpeakingPal.
“Boston-based Authess utilized machine learning and big data analytical technology to provide next-generation practical capability measurement test for enterprises in human resource recruitment to reduce the risk of hiring mismatched staff and to hire ‘responsive’ employees who can produce results at the business site.”
“Listenwise is an online listening service platform with strengths in non-fiction content actually broadcasted by National Public Radio (NPR) for use as learning material for schools.”
“Academic Merit develops services such as Fine Tune that help teachers write English composition teaching methods. University entrance examination in Japan is about to change dramatically and writing test will be introduced in 2020. Since it is not possible to clearly judge correct or incorrect answers in English composition writing exams like a mark sheet, it is important to do scoring with high precision without delaying the time schedule, and how to adjust the scoring criteria among graders becomes a problem to solve. Fine Tune is a service that aligns the scoring standards and scoring criteria of writing, practices scoring and gives feedback. Academic Merit also has a cloud services platform called Assessment 21 that students can learn English writing composition, and there are plenty of administrator functions that school teachers can use to manage progress.”
“ELSA is a venture from 500 Startups and has a technology that can evaluate pronunciation. English speaking learning content is developed based on that technology. There are two major reasons for investing in ELSA. The first is elemental technology of pronunciation evaluation. ELSA analyzes the pronunciation evaluation by adding the learners’ mother tongue elements based on deep learning to the pronunciation evaluation, and it has acquired user traction in a short time after releasing worldwide. For example, Japanese are not good at English pronunciation of R, but ELSA can recognize and evaluate the part affected by the learners’ mother tongue. The second point is that the pronunciation evaluation technology and learning service is tightly linked. The trend of evaluation in the education field is changing greatly. For example, formative evaluation is not evaluated at the end of the learning activity, it is done during the learning process to grasp the learning of the student. Evaluation alone will not be established as a learning service. It was good that ELSA’s service could respond to the learner’s consistent needs from evaluation to learning with reliable technology, as opposed to the subjective personal evaluation in speaking.”
“Israel’s CodeMonkey is a game-based learning platform for teaching children code. An AI engine simultaneously runs in the background and analyzes the players’ actions in order to deliver personal feedback, hints and instructions accordingly. During the game, students not only acquire the fundamentals of computer science, but 21st century skills, such as multistep problem solving and analytical thinking. CodeMonkey is also tailored to educators’ needs as it is designed as a curriculum that does not require prior coding knowledge and experience. CodeMonkey also provides basic training, management and reporting tools, automatic grading and interactive online training course (MOOC). CodeMonkey is adopted by over a quarter million students in over 2,000 schools nationwide in Israel, led by the Ministry of Education, Israel. In addition to Israel, it was adopted by educational institutions such as the United States, UK and France and China. We like investing in companies that have a committed founder who developed a sophisticated product that has a well-designed and user-friendly interface. CodeMonkey has developed incredible momentum, impressive technology, skilled arts and graphics via an expert team that is fundamentally changing the way children learn and use coding all over the world. On 6 Dec 2018, CodeMonkey has been acquired by China’s TAL Education for US$20m.”
“Israel’s SpeakingPal is an English speaking learning teaching material development startup company who has over 4 million users in 180 countries in the world. Edusoft, a subsidiary of US ETS with TOEIC® and TOEFL® and the world’s largest assessment body, is the major shareholder.”
“Looking ahead, we aim to differentiate itself from other EdTech venture investment funds by making maximum use of the management resources of the EduLab group, not only to pursue mere capital gain income, but also to secure profit by developing the products of our investee companies in other areas and markets, such as CodeMonkey sold in China.”
CEO Takamura summed up by emphasizing EduLab’s inspiring philosophy and mission as a roadmap for their development and growth journey in the fourth-generation testing technology era: “Does this test really measure individual ability? Carrying out testing has a purpose. However, seeking to measure something in line with this purpose is not a simple matter. Tests with many goals are possible, such as to measure the achievement level of a person’s ability, or the progress of their learning, or to find their relative position in a group of examinees, or to analyze the trends in a group of examinees. If the test isn’t designed correctly according to the purpose, it won’t be possible to perform accurate measurement. At EduLab, we think carefully how to measure human ability correctly, researching, developing and applying testing techniques to achieve that. Deviation doesn’t always indicate ability. The familiar deviation value is often misused. That is, it’s affected by the group of examinees. Besides the effects of the population, it’s also necessary to consider the difficulty of the test. It’s very strange if the difficulty of a test fluctuates each time, even though the pass mark remains fixed. One year might have a difficult test and thus very few successful candidates and another year might have a test with lots of relatively easy questions and thus a lot of successful candidates. Such a situation could result in inconsistent pass rates each year.”
“With item response theory, test difficulty and examinee ability are assessed separately. The probability of a person with a certain ability responding to a question of a certain difficulty is X%, and based on this, the ability of examinees is sought probabilistically. Since this requires extremely advanced statistics, it has still not been used widely in Japanese education, but overseas, it’s used widely for TOEFL® and other tests as a reliable method of measurement. The ability to establish a standard where a person of a certain ability can give correct answer to a certain question with a certain probability makes it possible to provide questions that adapt to the situation, similar to the approach used in eye tests, offering a slightly easier question if the examinee can’t answer a question, or offering slightly more challenging questions if the examinee correctly answers a question. Using a computer, the next question is varied according the examinee’s response. This computerized adaptive testing (CAT) enables highly accurate measurement of examinees.”
“Paper tests that are computerized (true score and deviation score tests) were the first generation. Adaptive testing based on item response theory was the second generation. Research is now underway on third generation testing called continuous measurement for children’s English tests and so on. Whereas earlier testing was confined to measurement, this approach actually supports learning. The items that must be acquired in a given unit are encoded in a mastery map on which testing is based. The test identifies areas that haven’t been completely learned, and through retesting or self-study, this approach aims to link measurement directly to efficient learning. EduLab is currently developing 4th generation AI technology for CASEC-GTS, CASEC-WT and so on. By disseminating and distributing the outcomes — accurate testing methods — we are contributing to effective education as well as to the development of individual ability.”