Overview
An AI-based Hybrid Chatbot Solution
-
An AI Hybrid Chatbot Solution
that's Even More Accurate! -
WISE iChat is an AI-based hybrid chatbot solution that comprehensively combines natural language processing technology which developed by itself for 24 years with machine learning, text mining, semantic analysis, search, etc.
Its feature is a hybrid-type chatbot conversation processing technology that includes machine learning-based accurate user intent analysis, iChat's unique five-way conversation processing, multi-turn conversation, Q&A knowledge-specific tools, webhook-style functional extension, etc.
-
Application of Wise Nut's original
core technology developed
in-house for 24 years -
Solutions with the largest
number of chatbot construction
cases in Korea -
Korea's best GS (Good Software)
first-grade certified product -
Product design based
on software process (SP)
certification
core technology
- Hybrid Conversation
Intent Recognition Technology -
To classify intentions of user questions, we applied a hybrid conversation
intent recognition technology that comprehensively combines machine learning,
pattern matching, similarity comparison, and sentence feature extraction based
on our own NLP technology.
- Conversation-Type Pattern-Based Knowledge Management
-
It supports accurate question answering and comprehensive knowledge
management by expanding conversational work through the separation
of five conversation patterns: immediate answer type for quick
responses, complex immediate answer type focusing on dialogue
keywords, user selection/branching type, operator-variable
conditional branching type, and additional query SLOT type
for refining answer searches.
- A highly scalable
conversational task structure
that can flexibly respond
to customer needs -
It supports various services that occur in conversational operations,
such as natural language understanding, knowledge management,
and conversation memory management. Each conversational operation
is organically connected to each other, but it is designed to be developed
and provided independently, so it consists of a conversational
task structure that can flexibly respond to customer needs.
- Conversation task
component Multi-turn conversation
technology -
To better identify and address the specific intents of user queries,
the Chatbot incorporates multi-turn conversation technology.
This includes features like handling repeated queries, branching discussions,
and posing follow-up questions to the user. It also supports transitioning
seamlessly to new conversations after concluding a current one and forwarding
relevant information when connecting related dialogues.
system diagram
-
A chatbot solution that can conversate
in specific areas through machine
learning-based user intent analysis
and various forms of conversation methods -
Wise iChat is a chatbot solution that can chat about specific areas through machine learning-based user intent analysis and various forms of conversation methods.
It is composed of logic that easily processes tasks such as Q&A, branching, and follow-up questioning by using asynchronous communication between the user and chatbot, leveraging external backbone systems outside the solution.
STRONG POINT
- Multi-turn Conversation Technology combining conversation task components
-
Existing functions, such as intent classification, entity recognition, etc., are defined as individual services used in conversation tasks.
It has a function that leads to another conversation after ending a conversation and a function that forwards information when connecting the conversation.
Logic can be configured for business processes such as repeating, branching, and questioning users in addition to question and answer.
- Improved accuracy of intent classification
-
Building knowledge optimized for the site through an intent classifier that combines machine learning, query similarity, and text patterns
Providing knowledge improvement tools for intermediate inspection, improvement work, and final accuracy evaluation of the constructed knowledge
An intent classifier optimized to handle multiple users simultaneously
- Webhook API method functional expansion
-
Entity recognition associated with site-specific custom modules, providing answers
Provision of linkage function to the backbone system
Provision of linkage function with already constructed statistical analysis module
Processing of real-time conversation history outside the solution
- Diversification of conversation management methods
-
Building knowledge optimized for the site through an intent classifier that combines machine learning, query similarity, and text patterns
Providing knowledge improvement tools for intermediate inspection, improvement work, and final accuracy evaluation of the constructed knowledge
An intent classifier optimized to handle multiple users simultaneously
REFERENCE
| Customer | Project | Year of construction | Application field | Project content |
|---|---|---|---|---|
 |
Civil complaint consultation chatbot 'Ara' | 2021 | Public service / Internal work |
|
 |
Conversational chatbot for civil complaint | 2021 | Public service |
|
 |
In-house work automation chatbot 'My KODI' | 2020 | Internal work |
|
 |
Construction of an internal chatbot | 2020 | Internal work |
|
 |
Data Mentoring System | 2020 | Public service |
|
 |
Myongji University Integrated Response Chatbot | 2020 | Public service |
|
 |
Financial Partners 'Aurora' | 2020 | Public service |
|
| Intelligent Communication Platform 'Molly' | 2020 | Internal work |
|
|
 |
Non face-to-face debt consultation service 'Saeromi' | 2020 | Public service |
|
 |
Civil complaint consultation chatbot 'Seoul Talk' | 2019 | Public service |
|
