These are talks presented by members of the community. Each new topic will feature an upcoming Community Talk.
Thursday, June 3, 2021 4:30 PM → Thursday, June 3, 2021 5:30 PM
Building on previous success in this area, the BioCorteX team have used TypeDB to map the therapeutic patent landscape providing unique insights and opportunities. We are able to establish the patent structure for potential therapeutics in a matter of seconds. Importantly, at BioCorteX we are able to quickly highlight the gaps that we refer to as the undiscovered country.
Nik Sharma is the CEO/Co-founder of BioCorteX. He is a clinician scientist at UCL with a specialist interest in neurodegenerative disease and the microbiome. Nik leads the first clinical trial of direct microbiome manipulation in people living with motor neuron disease (MND also known as ALS). The unique multidisciplinary team at BioCorteX combines expertise from neuroscience and aerospace with the explicit aim of developing a new approach to therapeutic optimisation. The four BioCorteX engines are purpose-built to develop enhanced therapeutics to address a range of disorders at scale. BioCorteX’s mission is to cure neurodegenerative diseases by hacking the microbiome and delivering enhanced therapeutics.
2021-06-08T16:30:00Z → 2021-06-08T17:30:00Z
Can we improve machines ability to reason by merging traditional knowledge graphs with causal graphs?
Jeff and his team are out to answer this question, using TypeDB. In this talk, Jeff will set out to cover:
- What is a domain knowledge graph?
- What is a causal graph?
- Why it is important to use both causal graphs and domain knowledge graph when building reasoning systems.
- How we are building a tool in node-red that will allow modellers to model knowledge graphs and causal graphs and the output of the model will be TypeDB schema (GQL)
Jeff Dagliesh, VP Product @ Geminos
Prior to joining Geminos, Jeff spent 18 years at Chevron. During the last 10 years he developed, and ran the worldwide drilling technology strategy, which included automating large parts of Chevrons worldwide drilling fleet, data integrations with geological and reservoir modelling systems.
While at Maana he built knowledge graphs for energy customers. At the end of his time at Maana he lead the team that built WellLine, a knowledge graph for oil wells that recreated the history of the wells based on running NLP over mining millions of documents and old reports about oil wells.
Today, his team is focused on building better reasoning systems for companies undergoing a transformation from reactive operations to predictive operations. They are looking to greatly improve how machine learning results can be explained, so the causes of the predictions can be controlled to create desired results. TypeDB is a place to instantiate the models and ask questions of the data in the models.
2021-06-09T16:30:00Z → 2021-06-09T17:30:00Z
Biological data has always been nuanced and deep. A scientist may spend their whole career probing just a small subset of this information. However, the utility of understanding biological data has become more imperative. The stakes are higher with the recent epidemic and rapid developments in biological technology.
In this talk we ask how we can use the open source knowledge graph to better understand viral biology in general?
- Biology is complicated and rich
- Deep scientific literature is accessible with the right tools
- Knowledge graphs are key to lowering the barrier to navigating complex domains
- Viral biology is varied and exciting and interesting in any context
- Viruses are on everyones mind, we can use this momentum to spread science
Principle Data Scientist
Gunnar Kleemann holds a PhD in Molecular Genetics from Albert Einstein College of Medicine and a Master’s in Data Science from UC Berkeley. He did post-doctoral research on the genomics of aging at Princeton University, where his research focused on developing high throughput robotic assays to understand how genetic changes alter lifespan and reproductive biology.
Gunnar is the Principal Data Scientist and owner of the Capital Data Corp of Austin (ACD) as well as the co-founder of the Berkeley Data Science Group. He is interested in how data science facilitates biological discovery and lowers the barrier to high-throughput research, particularly in small, independent labs. One of his passions is enabling scientific research through teaching, mentorship, and corporate engagements. In line with this goal, he offers consulting services to local businesses, teaches data science for the UC Berkeley School of Information, and runs pythonformakers.org.
Wednesday, June 23, 2021 4:30 PM → Wednesday, June 23, 2021 5:30 PM
Governments provide “opendata” about COVID19, such as PCR tests. With this opendata, we can juxtapose various periods of the past year (spring 2020 lockdown, progressive curfew in 2020 winter, then spring 2021 for departments suffering the greatest progression of the epidemic, holidays, fests… For each of these periods, either non-essential stores were closed, or the right to movement was reduced, the closures of bars, restaurants or sports venues, etc. Following these decisions, it was generally observed that the epidemic slowed down, then a more or less significant recovery following the reduction in measures.
In the face of the number of important and heterogeneous influencing factors, decision-making to fight this virus without penalizing the economy is in all cases difficult. Can we understand and do better at prevention? Precisely, how to objectify the influential parameters?
- Presentation of the problematic
- Definition of a generic graph model to be loaded in TypeDB and specialization by events and effects insights
- Mathematical modelization enrichment thanks to TypeDB consolidation and specific events identifying
- Relying on graph links between events and source of effects (lockdown, departments/countries cross-borders, weather, …)
- Opportunities of an opensource community to leverage analysis with other countries (datas service hosting, TypeDB IA contributions, …)
Jean-Paul MOCHET is a chief architect at Capgemini company with more than 20 years of experience in the integration of systems implementing data as a service architectures combining knowledge management, blockchain and / or modeling / simulation in various sectors (life sciences, ministry of defense, supply chain and Telecom). More particularly in the field of Health and Life Sciences, he has defined data architectures for digital health spaces, pharmaceutical groups or even health mutuals where the 360 ° patient is a major issue.
Thursday, July 22, 2021 4:30 PM → Thursday, July 22, 2021 5:30 PM
We are trying to understand diseases and what we can do to alleviate them.
Using TypeDB to build our knowledge graph, we face scale and complexity challenges due to the inherent complexity of the biomedical domain. Convinced that TypeDB offers all the modeling power needed to integrate data, infer facts, and predict node properties and edges, we have implemented a set of tools that address challenges with respect to data migration, analysis and visualization, as well as machine learning.
GraMi (TypeDB Migrator) allows for data to be migrated into TypeDB at scale while supporting the many possible ways to model schema. GraEs (TypeDB + ElasticSearch) allows for query auto-completion and improves graph exploration while providing a convenient backbone for more specific visualizations and analyses. Finally, we have collaborated with TypeDB to extend and generalize the existing KGCN (Knowledge Graph Convolutional Networks) library to be compatible with PyTorch Geometric, in order to meet your AI needs. All three are/will be open-sourced and made available to the community.
Henning, as a Sr. Computational Scientist at Bayer, has spent his career working with data infrastructures, and data analysis within the life sciences space. Today, Henning’s focus at Bayer is on drug discovery and disease understanding. Henning and team are building a knowledge graph to integrate a variety of data sources into one body of knowledge that can be reasoned over. TypeDB provides the database, and type system to empower his team’s work.
Tuesday, July 6, 2021 4:30 PM → Tuesday, July 6, 2021 5:30 PM
In this talk, Maydan explores and demonstrates various TypeDB capabilities by using content authorisation as an example.
Maydan has more than 10 years of experience in various roles including applied algorithmic research, software architecture, hands-on software development, and management of projects and people. Maydan has MSc. in industrial engineering with a specialization in machine learning which he has applied in a wide range of business domains including managing computational protein folding workflows, cell imaging analysis, ad tech algorithmic problems, and building cybersecurity solutions.
He is now founding his first company and building its main product. This led him to work with TypeDB’s unique abilities around the challenges of managing access to data and how to better manage a single source of knowledge. In his free time, Maydan likes trekking with his family, rock climbing, programming, and learning about new technologies with a special interest in biology.