Frequently Asked Questions
What are the data sources you use?
We build the Wisecube knowledge graph by parsing millions of abstracts and full text publications from MEDLINE™ and PubMed Central™, as well as other structured data sources such as WikiData, ClinicalTrials.gov, Uniprot, CHEMBL and many others. Additional proprietary content can be added to the platform through partnerships with publishers, or the internal document holdings of our Enterprise customers.
Do you process data from tables and illustrations?
How do you account for research that may not be accurate or reproducible?
Can you integrate 3rd party data sources such as patents, proprietary documents and databases?
Yes, we have integrated several 3rd party databases. New data sources are continuously being added to the platform. Please contact us if you would like to know what is planned next or would like to add a custom or proprietary data source. Also being an open platform, it is also possible for your data science team to use the Wisecube Knowledge Graph Engine pipeline to extend it.
Is my data secure with Wisecube Knowledge Graph Engine ?
How are you different than WikiData ?
In addition to incorporating the life sciences subset of WikiData, We also incorporate additional data from Clinicaltrials.gov, and inferred relationships from Pubmed. Even though there is some clinical trials data in WikiData, but it is often missing important elements, for e.g. links to the condition or intervention or links to outcomes.
Have you conducted any quality checks to ensure your Knowledge graph is reliable ?
Yes, we take quality very seriously. We ensure reliablity in multiple ways:
- We work with Wikidata and Semantic Data experts who have contributed a lot to WikiData who has come up with an extensive test set to validate the graph.
- We use existing sample queries used by WikiData to validate comparative quality.
- Finally, we compare our graph results to other knowledge graphs to ensure quality and reliablity.
How does information around clinical trials, conditions, and interventions get added to your Knowledge Graph?
We have built a custom parser that can extract specific information from clinical trials using state-of-the-art NER (Named Entity Recognition) models. We then fuse this with the underlying information from WikiData to provided augmented clinical trial entities.