FRONTEO Inc. (Headquarters: Minato-ku, Tokyo, President: Masahiro Morimoto, hereafter FRONTEO) is an in-house developed AI engine "Concept Encoder (trademark: conceptencoder, reading: concept encoder) that has achieved results in the life science AI business. As the second step of the AI ​​solution "WordSonar (trademark pending, reading: WordSonar)" that utilizes "" in the field of business intelligence, we analyze the "customer's voice" of financial institutions to detect signs of misconduct and customer loyalty. We have started to provide "Word Sonar for Voice View" (trademark pending, reading: Word Sonar for Voice View) that realizes improvement.

In recent years, in dealing with customers of financial institutions, the amount of data collected has increased dramatically while the conversion of call centers to DX (digital transformation) has led to the diversification and efficiency of communication methods such as telephone, email, SNS, and chatbots. Is increasing.Traditionally, the customer support department has been working on responding by classifying "customer voices" by methods such as people actually listening to some of the extracted voice data, viewing a list, and performing a keyword search. I did.However, due to the rapid increase in “customer feedback” data, it is difficult to grasp the overall picture in a short period of time and to decide and implement measures to enhance customer satisfaction.

WordSonar for VoiceView collects a huge amount of "customer voices" collected by various means such as text data of telephone calls at financial institutions, e-mails, chats, and daily reports written by counters and sales staff. It is possible to centrally aggregate and analyze as follows.

XNUMX. XNUMX.AI captures the characteristics of the collected text data and displays all the data on the map without bias.
By taking a bird's-eye view of the data displayed comprehensively on the map, it is possible to grasp the amount of inquiries and the positional relationship between the data. To prevent.In addition, by looking at the relationship between “customer feedback” and notable events such as “non-payment” and “cancellation”, inappropriate responses can be extracted.

Two.Timely understanding of "customer voice" trends through regular analysis
Observe and analyze changes in the quantity and positional relationship of each item of "Customer Voice" displayed on the map.By checking in chronological order, it is possible to catch signs of trouble or scandal, such as an increase or decrease in the number of inquiries due to the influence of products, seasonal factors, the external environment, etc.

From the analysis of "customer feedback," we can also capture positive reactions such as "the staff at the store responded well" and "the product description was easy to understand." WordSonar for VoiceView is a solution that realizes new service provision opportunities and improvement of customer loyalty by discovering and grasping not only problem improvement points but also results and deploying them to each department in the company as soon as possible.

WordSonar for VoiceViewAnalysis example by

If a financial institution makes a serious mistake or violates laws and regulations when dealing with customers, there is a risk of causing a situation that has a significant impact on business performance, such as administrative sanctions by regulatory agencies such as suspension of business, or deterioration of the corporate image.Taking life insurance companies in Japan as an example, the top 10 companies receive about 1 to 1 * complaints from customers every quarter.The Life Insurance Association of Japan has set up 4 categories of complaint items, and is working to realize customer-oriented business operations by grasping and clarifying the trends of complaints that occur.

* Source: The Life Insurance Association of Japan, life insurance company complaint reception information, insurance claims, etc. payment information
 https://www.seiho.or.jp/member/complaint/

Figure 1: <Analysis example> Visualization of "customer feedback" based on 37 complaint classifications by the Life Insurance Association of Japan

Using WordSonar for VoiceView, "customer voices" sent to life insurance companies are classified on the map for each of these 37 items.(Figure 1), Which items are most common, and because the items are close to each other and have similar properties, it is possible to check whether they are likely to occur at the same time.For example, it is easy to think that the items "insurance premium payment-related" and "cancellation procedure" have different characteristics when classified according to human senses. It understands the similarity of "refund" complaints and maps them closer together.You can also show the wording of specific inquiries that show similarities.(Figure 2).

Figure 2: The content of the complaint is similar between the "insurance premium payment relationship" and the "cancellation procedure" mapped nearby.

Furthermore, as a method of grasping the events that cause troubles and scandals in advance, by entering keywords such as "payment refusal", "non-payment of insurance money", and "dementia", highly relevant "customer voice" Can be highlighted(Figure 3)..For example, if you enter "dementia", "I was forced to take out a large amount of insurance as recommended by my elderly mother. I want you to cancel it because my judgment is weak." Even if you do not use the expression "dementia" such as "I was enrolled in new insurance", the "customer's voice" that AI has determined to be relevant is displayed in a list.In this way, troubles and scandals can be minimized by quickly grasping the exchanges that are suspected to be "inappropriate recruitment acts" that are overlooked by sample extraction or not detected by keyword search, and instructing the site to respond. It will be possible to suppress it.

Figure 3: Searching for "dementia" reveals highly relevant "customer voices"

Another feature of WordSonar for VoiceView is that it can analyze character strings that include ambiguous sentences or erroneous conversions.When analyzing "customer's voice" with conventional tools, there was a problem that it was not possible to properly extract sentences that contained even a small amount of typographical errors.However, since the AI ​​engine "Concept Encoder" analyzes words and sentences based on their meanings and usages, not just character strings, it is a text that includes notational fluctuations such as synonyms, and some erroneous conversions and typographical errors. However, it has been confirmed that the analysis result equivalent to the text with accurate description can be obtained.In addition, the "Concept Encoder Optimizer" (probability of risk factor occurrence), which was installed in the risk detection / prediction support system "WordSonar for AccidentView", which is the first in the "WordSonar" series, and was patented in January 1. We are also planning to implement a function) that clearly shows improvements and preventive measures to reduce the number.

FRONTEO will continue to apply the technology and know-how to analyze and evaluate text data using natural language AI to provide solutions that do not overlook risks and opportunities to meet the needs of customers in the financial business.

■FRONTEOabout　URL: https://www.fronteo.com/

FRONTEO uses the in-house developed AI engine "KIBIT", "Concept Encoder", and "Looca Cross" specializing in natural language processing to extract meaningful and important information from a huge amount of text data, and is a company. It is a data analysis company that supports the business of. Since its establishment in August 2003, it has been expanding globally to Japan, the United States, South Korea, and Taiwan, focusing on legal tech businesses such as "e-discovery (electronic discovery)" and "digital forensic investigation" that support corporate international litigation. Has been developed.Based on the AI ​​technology cultivated in this business, we will expand the business field to the life science field, business intelligence field, and economic security from 8, and by using AI to "turn text data into knowledge", We contribute to solving various corporate issues such as drug discovery support, dementia diagnosis support, financial, personnel, and sales support. Listed on TSE Mothers (currently TSE Growth) on June 2014, 2007. Obtained a first-class medical device manufacturing and sales business license in January 6 (permit number: 26B2021X1), and notified the managed medical device sales business in September of the same year (notification number: 13 Minato Misei Equipment No. 1).The capital is 10350 thousand yen (as of March 9, 3).

* FRONTEO, KIBIT, conceptencoder, and Coroban are registered trademarks of FRONTEO in Japan.