AI Start-up ExaWizards Launches “Qontextual,” FAQ Search Engine

2020/1/17 14:00

TOKYO, Jan. 17, 2020 /Kyodo JBN/ --

ExaWizards Inc.

- AI Immediately Learns Q&A with Excel Data, Automatically Analyzes It Based on Context of Question, and Gives Answers -

ExaWizards Inc. (hereinafter, “ExaWizards”), a Tokyo-based firm specializing in development of AI-enabled services for industrial innovation and social problems solutions, launched “Qontextual,” a FAQ search engine, on Thursday, January 16, 2020.

- Can be used for a wide range of purposes such as sophisticated search, call centers, chatbots, manual search, and counter operations.

Qontextual is an engine that solves various issues related to text, such as Q&A, text classification, and meaning search. It can be used for a wide range of operations, including sophistication of search within a website, support for call center and counter operations, accuracy improvement for chatbot answers, sophistication of search within in-house manuals, etc., and substitution of manuals for technical jobs.

By simply uploading a company’s Q&A data as Excel files, AI immediately learns it, automatically analyzes and searches answers best suited to the purpose of the search based on the context of the question, and it displays them in an appropriate order. This allows the company to introduce high-quality services in an easier and quicker manner than via traditional Q&A systems.

(Logo:
https://kyodonewsprwire.jp/prwfile/release/M105885/202001105604/_prw_PI1fl_PFhjC20j.jpg)

- Solves issues such as difficulty in finding appropriate answers and the huge workload required for introduction and maintenance.

Implementation of a Q&A system with a traditional language processing tool threw up issues such as difficulty in getting appropriate answers when searching with natural text as well as the huge workload required for introduction and maintenance, including category settings and classification of frequently asked questions.

With simple architecture and pre-learned natural language data, Qontextual solves multiple tasks in different data sets faster than traditional systems. Excelling at solving natural language tasks such as Q&A, text classification, and meaning search, Qontextual gives answers based on the context of the question in addition to words and also contributes to improving Q&A data. AI automatically analyzes and displays the sought answers based on search history, which reduces the workload required for introduction and maintenance, and makes business operations more efficient.

A paper by a Qontextual development engineer was selected by ACL, the authority in the field of natural language processing, in June 2019.

An Investigation of Transfer Learning-Based Sentiment Analysis in Japanese

Enkhbold Bataa: ExaWizards Inc./Tokyo, Japan

https://www.aclweb.org/anthology/P19-1458

Qontextual can be utilized for any natural language processing tasks, not only sentiment analysis adopted in the paper.

Overview of Qontextual services

Features:

(1) Able to give answers based on the context of the question

- Able to answer searches with natural text thanks to the general-purpose language expression model based on deep learning

- AI automatically searches answers best suited to the purpose of the search based on the context of the question and gives answers in an appropriate order

(2) Able to learn and re-learn Q&A data faster

- Able to immediately vectorize and utilize Q&A data

- Users only need to upload data in Excel or CSV format

(3) Equipped with a reporting function

- Examines what kind of searches are actually made based on similar contexts and shows the ranking

(4) Multiple languages support

- Able to respond to questions in multiple languages, including Japanese, English, Chinese and others.

Applicable cases:

- Sophistication of search within a website

- Support for call center operations

- Support for counter operations

- Accuracy improvement for chatbot answers