Two talks: KGs in ArangoDB, and KGs for Entity Disambiguation

Knowledge Graph team at Columbia welcomes the tech community to join us for an evening on talks on application and construction of Knowledge Graphs! We are fortunate to have speakers from Bloomberg LP, Dr Maria Pershina who will give a talk on Entity Disambiguation using Knowledge Graphs. She will be followed by Head of Engineer, Dr Jörg Schad, who will provide a practical example on creating a knowledge graph using ArangoDB.

We will have plenty of opportunity for networking, Q&A and getting deep insight on how to build an effective Knowledge Graphs and its vast applications. The event will take place at 308A Lewisohn Hall at Columbia University.

P.S: The audience is not expected to have any prior experience or cognizance of Knowledge Graphs.

Schedule
7:00 PM - Doors Open
7:10 PM - Knowledge Graphs for Entity Disambiguation
7:40 PM - Q&A
7:50 PM - Building Adaptive Knowledge Graphs
8:20 PM - Q&A
8:30 PM - Event Ends

Talks

1. Knowledge Graphs for Entity Disambiguation - Dr Maria Pershina
   Entity Matching is the problem of determining if two entities in a data set refer to the same real-world object. In the last decade, a growing number of large-scale knowledge bases have been created online. Tools for automatically aligning these sources would make it possible to unify them in a structured knowledge and to answer complex queries.

The task of Named Entity Disambiguation is to map entity mentions in the document to their correct entries in some knowledge base. NED is both useful on its own and serves as a valuable component in larger Knowledge Base Construction systems. In this talk, we will show how to construct a Knowledge Graph for the above problems and how to adapt Personalized PageRank to resolve entity ambiguity for both tasks. Our approach is domain-independent, robust to incomplete data, scalable, and does not require training and probabilistic inference.

2. Building Adaptive Knowledge Graphs - Dr Jörg Schad
   Information within organizations cover various topical domains and knowledge graph projects seek to make all this knowledge accessible. Modern knowledge graphs have to be adaptive and allow for an easy combination of various aspects of information an organization produces – best case, in an ad-hoc fashion.

A combination of JSON stores, semantic search and graph technology is often used to provide native storage and access to data. But these setups tend to be challenging in terms of data consistency and development complexity – therefore hindering the adaptiveness of knowledge graph projects.

In this talk, Jörg will show how to use a native multi-model database to create a knowledge graph over movie data scraped from Wikipedia by using the graph and semantic search capabilities within ArangoDB. Different aspects of movie metadata can be represented as a graph to essentially process textual information into main components and their relationships. He will introduce the new features in ArangoDB, that can play a crucial role in helping users to access the substantial and valuable knowledge from this knowledge graph. He will also show the possibilities to enrich this knowledge further with e.g. plot summaries for building ML engines like Recommendation system, Text Classification etc.

About the Speakers
Dr Pershina works as a research engineer at Bloomberg LP in the computational intelligence team. Previously, she has worked at Goldman Sachs, Microsoft and Google. She holds a PhD degree from NYU where she researched in the crucial areas of Knowledge graphs such as entity disambiguation, entity linking, distant-supervision based relation extractions.

Jörg Schad is Head of Machine Learning at ArangoDB. In a previous life, he has worked on or built machine learning pipelines in healthcare, distributed systems at Mesosphere, and in-memory databases. He received his PhD for research around distributed databases and data analytics. He’s a frequent speaker at meetups, international conferences, and lecture halls.