Symbolic Knowledge Discovery

Objectives

Knowledge discovery in databases (KDD) has emerged as major field of research and applications, due to the ever increasing creation of data in fields of great societal impact, such as natural sciences, healthcare, banking, marketing, the world wide web, etc. KDD consists of processing possibly large volumes of data in order to discover meaningful patterns (knowledge units). These patterns can be interpreted and used for decision making purposes, and reused for refining the discovery process. KDD is usually based on three main steps: data preparation, data mining, and interpretation of the discovered patterns (descriptions). Symbolic knowledge discovery (SKD) introduces students to a knowledge discovery framework with strong mathematical foundations and based on conceptual thinking and knowledge processing and reasoning. It is intended for mathematics, data science and computer science masters’ students with interests in data analysis and symbolic machine learning.

Syllabus

Part I - Introduction to knowledge discovery and mathematical background: Chapter 0 - Brief introduction to knowledge discovery. Chapter 1 - Basic mathematical background. Part II - Formal concept analysis, concept exploration and implications: Chapter 2 - Introduction to Formal Concept Analysis (FCA). Chapter 3 - Main algorithms of pattern mining with FCA. Chapter 4 - Iceberg lattices: Motivation, interestingness measures, working examples Part III - Refinements, Pattern structures and Applications: Chapter 5 - Association rules. Chapter 6 - FCA in classification. Chapter 7 - Further topics: Multimodal clustering (biclustering and examples), From FCA to biclustering, clustering complex data, pattern structures, from FCA to pattern structures,further applications (recommendation, text mining, information retrieval, etc.)

Prerequisites

Symbolic knowledge discovery is expected to be self-contained but some elementary background in discrete mathematics, order theory and computational complexity is recommended.

Cross Competence Component

The course unit fosters the development of transversal skills in critical thinking, creativity, and problem solving strategies, both in class, through independent work, and across the various assessment components.

Laboratorial Component

A selection of 5 to 10 exercises will be provided to students every week. These will be partially solved in class and the remaining exercises require students’ personal work. Also, students are expected to actively contribute to their respective groups towards their group’s final project.

Programming And Computing Component

Some programing in Python will be expected from students towards the completion of their final project.

Ethical Principles

All member of a group are responsible for the group’s work. In any assessment, every student shall honestly disclose any help received and sources used. In an oral assessment, every student shall be able to present and answer questions about the entire assignment and solution.

Periodic problem sets, final exam and project.

Bibliografia Principal

• Introduction to Formal Concept Analysis: R. Belohlavek 2008 Palacky University, Olomouc
  • Concept Data Analysis: Theory and Applications: C. Carpineto, G. Romano 2004 John Wiley Sons
  • Finite Ordered Sets: Concepts, Results and Uses: N. Caspard, B. Leclerc, B. Monjardet 2012 Cambridge University Press
  • Introduction to Lattices and Order: B.A. Davey and H.A. Priestley 2002 Cambridge University Press, (2nd ed.).
  • Conceptual Exploration: B. Ganter, S. Obiedkov 2016 Springer
  • Formal Concept Analysis: B. Ganter, R. Wille 1999 Springer-Verlag Berlin Heidelberg