Knowledge Engineering and Business Intelligence

Teacher:

• Knut Hinkelmann
• Holger Wache
• Nadeem Qaisar Mehmood

Lessons schedule: The course will be held in the following periods:

• April 14th - 16th
• April 28th - 30th
• May 5th - 7th
• May 19th - 21st

In addition there will some tutoring and laboratory hours. For the days April 14th-16th the timetable is the following:

• Tuesday - April 14th, from 10am to 1pm (room: Tim Berners-Lee)
• Wednesday - April 15th, from 2pm to 5pm (room: Tim Berners-Lee)
• Thursday - April 16th, from 9am to 12pm (room: Turing)

For the second week, April 28th-30th the proposed timetable is the following:

• Tuesday - April 28th, from 10am to 1pm (room: Tim Berners-Lee)
• Wednesday - April 29th, from 2pm to 5pm (room: Tim Berners-Lee)
• Thursday - April 30th, from 10am to 1pm (room: Turing)

For the third week, May 5th-7th the proposed timetable is the following:

• Tuesday - May 5th, from 10am to 1pm (room: Tim Berners-Lee)
• Wednesday - May 6th, from 2pm to 5pm (room: Turing)

Knowledge-intensive processes are more unstructured processes with a lot of involvements of users with their experience. These users need experience and knowledge at different levels for their work and decision making. Supporting such processes at their levels requires modelling and enacting several different forms of knowledge.

For connecting people to other knowledgeable people or providing access to relevant information mainly structural knowledge is used as meta-knowledge to support indexing and intelligent information retrieval. For preprocessing information etc. and even automatically derive and suggest possible solutions the knowledge itself has to be formally represented in order to allow automated reasoning.

For example, rule systems may support users during their decisions; knowledge can be used to define explicitly the goals and precondition of one task in order to constrain the process control flow etc. In general more explicit represented knowledge allows better support. But different forms of knowledge need different intuitive and adequate representations and inferences.

After completion of this module, the participants will be able to assess which kind of knowledge representation and reasoning is adequate and are able to develop appropriate knowledge-based systems. They can value the advantages of knowledge-based systems with respect to their costs.

Business Intelligence is concerned with supporting business decisions with facts. It supports business actors in turning data into knowledge that helps to make the right decisions. The module looks at different kinds of decisions (and hence requirements), at different kinds of data and different kinds of tools required to distill knowledge out of data.

This module is concerned with the acquisition, the representation and the inference of knowledge. Various forms of knowledge representation and inference techniques are analysed and discussed in detail. This course will include topics like

• Introduction: Knowledge in processes
• Decision Tables
• Rules
  • Textual represented rule (i.e. Horn clauses)
  • Forward and backward chaining
  • Data-driven and Goal-oriented
  • Negation-as-failure
• Object-centred Systems
  • F-Logic
• Fuzzy Logic
• Learning from observations
  • Supervised Learning
  • Classification
    • Decision Trees
• Introduction into Business Intelligence
• Business Performance Management
• Multidimensional modeling
• Data Warehousing
• Data Mining

Course Material

• Organisation

• Lecture “Introduction”
  • Slides
  • Exercise: Knowledge-intensive tasks

• Lecture “Rule-based-Systems: Decision Tables”
  • Slides
  • Exercise: Build a descision model with tables (incl. solution)
    • Process and Decision Models for Admission Process

• Lecture “Rule-based-Systems: Logic Programming”
  • Slides
  • Reasoning example: Ancestor
  • Exercise: Family example in Prolog with Solution
  • Exercise: Mini-Sudoku with Solution (Sources)
  • Home Exercise: Admission Process supported by rules
    • Solution

• Lecture “Forward and Backward Chaining”
  • Slides

• Lecture “ObjectLogic”
  • Slides
  • Exercise: family example with solution
  • Home Exercise: Admission Process supported by ObjectLogic

• Lecture “Fuzzy Logic”
  • Slides
  • Exercise: Fuzzy Sets (Solution)
  • Exercise: Fuzzy Set Operations (Solution)
  • Home Exercise: Credit Analysis

• Lecture “Learning from Observations”
  • Slides
  • Reading Material
  • Exercise: Decision Trees

* Lecture “WEKA Introductory Live Session”

• Intro

Software and Tools

• BPCDMN - Modeling tool for business processes, case management and decisions
• DMN Decision Table Template
• SWI-Prolog

Exam Date A.Y. 2014/2015

• Summer session: June 29th, 14:00-17:00

Exam rules:

• The exam comes in two parts
  • a course assignment (the data sheet (please unzip))
  • oral exam held in Olten, Switzerland, on 29th June 2015. (If you can not attend to that oral exam please contact your lecturer immediately!)
• The course assignment can be done in groups with 3 persons as maximum.
• The oral exam will be individualy and will last at least 45 Minutes per student.
• Both parts need to be passed separately. You can NOT use one part to balance the other part!
• The grade will be a combination of the grades of both parts. The ratio is 30% assignment and 70% oral exam.

Exam Results

• will be told when oral exam is finished