Interfacing Chatbot with Data
Retrieval and Analytics Queries for
Decision Making
Gan Keng Hoon, Loo Ji Yong and Goh Kau Yang
School of Computer Sciences
Universiti Sains Malaysia
The 6th International Conference on
Robot Intelligence Technology and
Applications 2018
Pullman Lakeside Hotel,
Putrajaya, Malaysia
Dec 16, 2018 - Dec 18, 2018
The Softer Side of A Bot
A Chatbot
Chatbot for Decision Makings
• Ordering Food
• Book a Ticket
• Get Health Tips
• Get Makeup Tips
& …
Task Oriented Chatbot
Master a single domain, e.g. knowledge
about a company, a domain, a skill scope
etc.
Able to have conversational interactions
and context awareness.
Can have some element of personality.
Chatbot ModellingTask-basedChatbot
Modelling
Conversation
Interactions
Domain Specific Task
Conversation Contents
Static Contents
Classified Responses
Simple Retrieval
Dynamic Contents
API/Functions
Structured Query
Data Retrieval
Decision Making
Analytics Queries
What is the total articles published in the 1st quarter?
Who published the most?
What are the popular research areas?
Please compare the citations for year 2017 and the year before.
Is there a correlation between the
number of articles published and
the number of supervised students?
Analytics
Chatbot
“total articles published in the 1st quarter?"
// generate a corresponding intermediate object with
the following contents
intent: count (from keyword “total")
target entity: article
constraint entity: date
constraint entity values: january, february, march
// from the object, SQL processing engine will
generate a query as following
SELECT article’, COUNT(article') FROM
PUBLICATIONS
WHERE ‘date' = “jan” or ‘date' = “feb” or
‘date’=“mar”;
Interface
Conversational-based Natural Language
Understanding
“total articles published in the 1st quarter?"
// generate a corresponding intermediate object with the
following contents
intent: count (from keyword “total")
target entity: article
constraint entity: date
constraint entity values: january, february, march
Text Preprocessing
Natural Language
Processing
Text Normalization
Information
Extraction
Intent to Business Logic Translation
// generate a corresponding intermediate object
with the following contents
intent: count (from keyword “total")
target entity: article
constraint entity: date
constraint entity values: january, february, march
// from the object, SQL processing engine
will generate a query as following
SELECT article’, COUNT(article') FROM
PUBLICATIONS
WHERE ‘date' = “jan” or ‘date' = “feb” or
‘date’=“mar”;
Actionable Data
Interface Mapping
Structured Query
Formulation
Function/API
Calls
Natural Language
Generation
Data Retrieval & Analytics based
Conversational Framework
Nybble The Robotic Kitten
Questions/Discussions
THANK YOU
Visit our ‘NOT QUITE ROBOTIC’ works at ir.cs.usm.my
Drop me an email at khgan@usm.my

Interfacing Chatbot with Data Retrieval and Analytics Queries for Decision Making

  • 1.
    Interfacing Chatbot withData Retrieval and Analytics Queries for Decision Making Gan Keng Hoon, Loo Ji Yong and Goh Kau Yang School of Computer Sciences Universiti Sains Malaysia The 6th International Conference on Robot Intelligence Technology and Applications 2018 Pullman Lakeside Hotel, Putrajaya, Malaysia Dec 16, 2018 - Dec 18, 2018
  • 2.
  • 3.
  • 6.
    Chatbot for DecisionMakings • Ordering Food • Book a Ticket • Get Health Tips • Get Makeup Tips & …
  • 7.
    Task Oriented Chatbot Mastera single domain, e.g. knowledge about a company, a domain, a skill scope etc. Able to have conversational interactions and context awareness. Can have some element of personality.
  • 8.
    Chatbot ModellingTask-basedChatbot Modelling Conversation Interactions Domain SpecificTask Conversation Contents Static Contents Classified Responses Simple Retrieval Dynamic Contents API/Functions Structured Query
  • 9.
  • 10.
  • 11.
    Analytics Queries What isthe total articles published in the 1st quarter? Who published the most? What are the popular research areas? Please compare the citations for year 2017 and the year before. Is there a correlation between the number of articles published and the number of supervised students?
  • 12.
  • 13.
    “total articles publishedin the 1st quarter?" // generate a corresponding intermediate object with the following contents intent: count (from keyword “total") target entity: article constraint entity: date constraint entity values: january, february, march // from the object, SQL processing engine will generate a query as following SELECT article’, COUNT(article') FROM PUBLICATIONS WHERE ‘date' = “jan” or ‘date' = “feb” or ‘date’=“mar”; Interface
  • 14.
    Conversational-based Natural Language Understanding “totalarticles published in the 1st quarter?" // generate a corresponding intermediate object with the following contents intent: count (from keyword “total") target entity: article constraint entity: date constraint entity values: january, february, march Text Preprocessing Natural Language Processing Text Normalization Information Extraction
  • 15.
    Intent to BusinessLogic Translation // generate a corresponding intermediate object with the following contents intent: count (from keyword “total") target entity: article constraint entity: date constraint entity values: january, february, march // from the object, SQL processing engine will generate a query as following SELECT article’, COUNT(article') FROM PUBLICATIONS WHERE ‘date' = “jan” or ‘date' = “feb” or ‘date’=“mar”; Actionable Data Interface Mapping Structured Query Formulation Function/API Calls Natural Language Generation
  • 16.
    Data Retrieval &Analytics based Conversational Framework
  • 19.
  • 20.
    Questions/Discussions THANK YOU Visit our‘NOT QUITE ROBOTIC’ works at ir.cs.usm.my Drop me an email at khgan@usm.my