|
RefViz uses mathematical algorithms to divide a set of papers into concept-based groups. Starting with the vocabulary determined from the title and abstract/notes fields for each set of references, RefViz uses a statistical model to find key concepts. Unlike other literature sorting applications that use predefined rules, RefViz defines key themes based on the context of the entire set being evaluated.
RefViz sorts references into groups in the following steps:
1. Identifies words that represent key concepts, primary and secondary, for a reference set.
2. Uses the key concepts to create a mathematical signature for each document.
3. Applies standard mathematical clustering algorithms to partition the document set into groups of similar papers.
The result is a much richer comparative analysis than a simple categorisation of documents based on word counts. Coupling this with powerful visualisation methods provides rapid insight into large reference collections.
|
