Probabilistic Knowledge Assembly Framework
This server provides a Web frontend to the ProbAbilistic kNowleDge Assembly (PANDA) framework and repository developed in the context of the DARPA's Big Mechanism research program.
What are Big Mechanisms?
"Big mechanisms are large, explanatory models of complicated systems in which interactions have important causal effects. The Big Mechanism program aims to develop technology to read research abstracts and papers to extract pieces of causal mechanisms, assemble these pieces into more complete causal models, and reason over these models to produce explanations. The domain of the program is cancer biology with an emphasis on signaling pathways."
What is knowledge assembly for Big Mechanism?
"Given a prior big mechanism and many fragments, the job of assembly is to extend the mechanism as thoroughly as is warranted by the fragments and other knowledge in databases and ontologies. Assembly also involves finding semantic inconsistencies among fragments, and finding “holes” or parts of big mechanisms for which no causal fragments are known."For more details see a reference document (PDF).
PANDA frameworkThe ProbAbilistic kNowleDge Assembly (PANDA) framework is aimed at supporting the assembly phase of the Big Mechanism. It is based on the concept of augmenting a semantic representation model with probabilistic learning and inference:
Knowledge Assembly for Big Mechanism
This repository contains RDF data with the information extracted from a selected corpus of PubMed papers. This data is further annotated with probabilistic information, partially generated using the ProbLog system. The model is grounded in the Probabilistic Knowledge Assembly Ontology & Uncertainty Ontology. All data artifacts can be directly downloaded or queried using the live SPARQL endpoint featured by this server.
- Datasets and ontologies (HTML)
- Extracted statements (HTML, CSV)
- Assembled events with uncertainty scores and provenance details:
- Assembled events with uncertainty scores and structure details:
- Assembled events with uncertainty scores and explicitly marked white- and black-listed chemicals (HTML, CSV)
- The initial model (HTML)
- Events corroborating the initial model (HTML, CSV)
- Events contradicting the initial model (HTML, CSV)
- Events extending the initial model (one common participant) (HTML, CSV)
- Missing event participants suggestions (HTML, CSV)
The code and data models for populating and managing this repository is available at GitHub.
Further information and contactThis repository is built on top of OpenLink Virtuoso Universal Server open source edition.
The list of contributors:
- Brunel University London: Szymon Klarman, Larisa Soldatova.
- NaCTeM, University of Manchester: Sophia Ananiadou, Riza Batista-Navarro, Chryssa Zerva.
- University of Manchester: Ross King, Robert Stevens, Martin Carpenter, Katherine Roper.
- Past contributors: Raheel Nawaz.