KBank joins with CU and NECTEC to develop NLP program to aid Thai-language interaction with computers

Wednesday 11 December 2019 16:19
KASIKORNBANK (KBank) has partnered with Chulalongkorn University (CU), the National Electronics and Computer Technology Center (NECTEC) and KASIKORN Business-Technology Group (KBTG) to conduct research on natural language processing (NLP), a branch of artificial intelligence (AI) that can facilitate interactions between computers and humans using natural language. The research aims to allow computers to learn and analyze complex languages, from single words to full sentences, and communicating messages just as humans do. It is hoped that the program will serve as a prototype for government agencies and private organizations to be applied for the nation's benefit.

Ms. Kattiya Indaravijaya, KBank President, revealed that KBank, in collaboration with CU, NECTEC and KBTG, has developed a Thai NLP program – an innovation that can process Thai language used in the banking and business sectors. The NLP technology helps computers to understand our natural language accurately and swiftly. Its main components involve language comprehension and computing power. CU is famous for its personnel with competency in both linguistics and computers, while NECTEC is well-versed in AI and has conducted extensive research in this technology. Meanwhile, KBank, a leader in the financial and banking sector, has contributed to this program through data support, thus ensuring that the research is practical and suited to the business sector. The cooperation of these parties has ensured the program's successful conception. KBank has done thorough program testing and has seen obvious progress, as it can now understand Thai language more deeply and accurately. The Thai NLP will serve as a model for further development and application to banking business or other organizations.

KBank uses Thai NLP, a natural language processing innovation, to enhance the ability of computers to understand messages that customers enter into various systems, to provide services and answer questions through chatbot via social media channels more accurately and quickly. The program also analyzes results to gain customer insights based on their collected comments. In the past, KBank had to hire other companies to conduct such research. This outsourcing tended to be expensive – costing several millions of Baht – and it could process only a small volume of customers' opinions. KBank is now able to extensively collect the customers' comments, covering KBank's base through various channels within a short period of time, and uses such information to develop products or services catering to the target customers' preferences. An example of overseas businesses that have applied NLP in their operations includes some e-marketplaces in China, which use NLP to analyze what their customers have spoken, written and read as the components to present products that match the customers' desires, or to use chatbots instead of humans to introduce products and after-sale services.

Once Thai NLP is developed to become a "smart" computer program, KBank will have a virtual assistant readily available to assist our customers with services that surpass their expectations and are truly fulfilling. In addition, we think that NLP will change the business operation patterns of commercial banks and other businesses tremendously by supporting analytics, documents and knowledge management of KBank, together with screening candidates applying to work for KBank. The system will also summarize essential information from loan application documents for loan approval, and in the future it could be used to verify documents, contracts and transactions.

Professor Bundhit Eua-arporn, Ph.D., President of CU, said that Thai NLP research is a big challenge when compared to communication in English. The pattern of Thai language is complex because it is a tonal language with vowels having different connotations. Moreover, there are always newly-invented words in accordance with different functions and different periods. In particular, the Thai language used in social media, such as the linguistics in finance and banking, is different from other business languages. Moreover, the formal language used in government is different from the language used in daily life.

Prof. Bundhit added that CU has a long-standing interest in Thai NLP research. At present, interest in Thai NLP and AI is growing as part of a similar trend worldwide. Despite this surge of interest, the number of Thai researchers in this field remains small compared to other countries. CU therefore aims to nurture interest in this field among new researchers. CU now teaches NLP courses at the Faculty of Arts and Faculty of Engineering. The new generation of lecturers is playing a pivotal role in research and development. KBank also provides scholarships to CU students and allows them to use the information and Open API obtained from this research. CU views that the development process must be systematic and sustainable, while the research must be able to help personnel gain the skills to respond to volatility in the new era and be integrated with the industrial sector. For this reason, CU has established the School of Integrated Innovation, with the mission to foster a new generation of personnel with the necessary skills to create innovations via new technologies. CU has also established a University of Technology Center, with the task of connecting CU's researchers with those in the industrial sector so that they can together conduct in-depth research in technological innovations that are of benefit to Thai society. The CU Enterprise unit ensures that innovations derived from research are developed into products in the form of startups. These three units will serve as important mechanisms in spurring innovations for national development, going forward.

Dr. Chai Wutiwiwatchai, Executive Director of NECTEC, said, "NECTEC has been established for more than three decades, and we have always placed importance on Thai NLP. This cooperation is an important milestone between private and public sector organizations to develop Thai NLP into an innovation that goes hand-in-hand with AI. The banking sector is one of the business groups that applies Thai NLP in its operations due to the diversity of data involved and the fact that it has to offer the most satisfying service experience for customers via accurate and complete data sent at an ever-faster pace. NECTEC has joined this partnership in its capacity as a national research unit with the policy and mission to support research with results that are applicable. "

NECTEC recently unveiled an AI service platform, AI For Thai, to facilitate the processing of Thai language, phonetics and images. AI For Thai can be used to make sentiment analysis, create a chatbot, translate speech to text and vice versa, and identify faces and objects. Programmers are welcome to test this service platform or join us in developing new services for AI For Thai. The AI-related knowledge gained through the joint research of KBTG and NECTEC will be developed for public benefit via AI For Thai, with the aim of promoting the increased use of AI in Thailand from now on.