Introduction

This seminar deals with different aspects related to Knowledge Graphs. Ranging from Linked Data to applying data mining techniques on Knowledge Graphs

Knowledge Graphs are large graphs used to capture information about the real world in such a way that is useful for applications. In these data structures, there are all sorts of entities (for example, people, events, places, organizations, etc.). These graph also contains all sorts of information about these entities (e.g., age, opening hours, …) and relations between them (e.g., this shop is located in Aachen).

Knowledge Graphs are used by many organizations to represent the information they need for their operations. The most well-known is Google, where a knowledge graph is used to enrich the search results. Also personal assistants, like Amazon’s Alexa, Apple’s Siri and Google Now, as well as question answer systems like IBM Watson, make use of knowledge graphs to provide information to their users. Besides these, also other information graphs, are in use by large organizations to improve or personalize their services. Examples include the Facebook graph, the Amazon product graph, and the Thompson Reuters Knowledge Graph.

One effort to create a world-wide Knowledge Graph is Linked Data. The World Wide Web made it possible to exchange documents and services on a global level - one can access and display documents from the other side of planet instantaneous, without prior agreement. Linked Data aims to achieve the same for data - to make data accessible, usable, queryable regardless from where the data is coming from or what the contents is. Linked Data does not replace the current Web. It adds instead an interoperable data layer based best practices, on open standards and technical components which define how data should be published and interlinked. The purpose of this seminar is to provide a conceptual and technical introduction to Linked Data and discuss individual approaches as well as state-of-the-art. An understanding of the basic concepts will then make it possible to discuss opportunities and challenges of Linked Data.

To make use of the data contained in the graph, one often needs to extract information from it. This is not always trivial as the data in the graph can be varying, noisy, dynamic, and usually there is a vast amount of it. So, specific techniques need to be used to be able to cope, specifically when one wants to use the graph data as an input for machine learning.

Preparing for the course

One options to prepare yourself for the course is watching trough these videos by Harald Sack: https://open.hpi.de/courses/semweb2015

A former youtube/google researcher published a presentation on different techniques used for search and discovery: YouTube

Preliminary schedule

TBD

Required Prior knowledge

We expect that you are able to read papers in which data structures and algorithms are used. Having experience with graphs is not required, but is definitely useful.

Note that the language of this course is English.