PubRunner

PubRunner automates periodic retrieval and processing of PubMed abstracts to maintain up-to-date biomedical text-mining corpora for analyses such as word2vec model updating and literature-based hypothesis generation.

Key Features:

• Automated Scheduling: Automatically retrieves the most recent PubMed abstracts on a user-defined schedule.
• Integration with Text Mining Tools: Incorporates user-specified text mining tools into the processing workflow for customized analyses.
• Data Dissemination: Uploads processed results to public repositories such as FTP or Zenodo datasets.
• Publicization of Results: Publishes the location of processed data on the PubRunner website.

Scientific Applications:

• Up-to-date corpora for text mining: Maintains current literature corpora to support reproducible and contemporaneous text-mining analyses.
• Improved understanding of biological interactions: Enables analysis of the latest findings to refine interpretation of complex biological interactions.
• Faster hypothesis generation: Provides current literature-based evidence to accelerate generation of experimentally testable hypotheses.
• word2vec re-training: Re-runs word2vec on the latest PubMed abstracts to generate current biomedical word vector representations.

Methodology:

Automated retrieval of recent PubMed abstracts on a user-defined schedule; processing of abstracts with a user-specified text mining tool; uploading processed data to public repositories such as FTP or Zenodo and publishing the data location on the PubRunner website.