Course timings: Mon, Thu. 6pm-7:30pm

Overview

Knowledge graphs are graph-representations used to capture information about real-world so that we can build applications that can use this knowledge effectively. Typically these graphs represent information about millions of entities (e.g., people, organizations, places, events, things,…) and model not only the relationships between these entities but also capture information about them (date of birth, founding date, opening hours, service dates, alternate names,…). Such knowledge graphs are used by many organizations to drive their applications – e.g., Google’s search engine uses their planet-scale Knowledge Graph, Microsoft has a similar knowledge graph to drive many of their applications including conversation systems, Amazon has a product knowledge-graph, IBM’s Watson uses combination of multiple knowledge graphs for completing complex tasks, National Institutes of Health (NIH) in US have a knowledge graph of medicines and related information, and many more.

(Tentative) Course Plan

In this course, we will focus on data management issues emanating while serving applications based on massive scale knowledge graphs. Currently the tentative topics that we plan to cover in the course include:

• Introduction to Knowledge Graphs
  • Graph representation of Information
  • Entity-relationship Graphs
  • Type information and Ontology
• Modeling Knowledge Graphs
  • RDF/OWL
  • Property Graph Model
  • Modeling complex knowledge - time-series, geo-spatial-temporal etc.
• Use of Knowledge Graphs in Information Retrieval
  • Use of KG Embeddings for complex retrieval tasks in IR
• Storage and Querying of Knowledge Graphs
  • Complexity of querying in Knowledge Graphs
• Recent Trends in Knowledge Graphs in IR

Background required

• Database Systems (equivalent of COL760 or COL762) necessary
• Adv. Data Structures (equivalent of COL702) preferred
• Logic (COL765 or COL703) desirable

Slides

1. Course slides are available here (until the end of the semester)
2. We also used slides from the Knowledge Graphs course by Prof. Markus Kroetzsch at TU Dresden.
3. For the use of knowledge graphs in Information Retrieval, the resource is the tutorial on Utilizing Knowledge Graphs in Text-Centric Information Retrieval by Laura Dietz, Alexander Kotov, and Edgar Meij. (Github).