Support Vector Machine


  • A state-of-the-art Supervised Learning Algorithm that separates Relevant from Non-Relevant Documents using geometric methods (i.e., geometry). Each Document is considered to be a point in [hyper]space, whose coordinates are determined from the Features contained in the Document. The Support Vector Machine finds a [hyper]plane that best separates Relevant from Non-Relevant Training Examples. Documents outside the Training Set (i.e., uncoded Documents from the Document Collection) are then Classified as Relevant or not, depending on which side of the [hyper]plane they fall on. Although a Support Vector Machine does not calculate a Probability of Relevance, one may infer that the Classification of Documents closer to the [hyper]plane is less certain than for those that are far from the [hyper]plane. 1
  • A machine-learning approach, used for categorizing data. The goal of the SVM is to learn the boundaries that separate two or more classes of objects. Given a set of already categorized training examples, an SVM training algorithm identifies the differences between the examples of each training category and can then apply similar criteria to distinguishing future examples. 2 3


  1. Maura R. Grossman and Gordon V. Cormack, EDRM page & The Grossman-Cormack Glossary of Technology-Assisted Review, with Foreword by John M. Facciola, U.S. Magistrate Judge2013 Fed. Cts. L. Rev. 7 (January 2013).
  2. Herb Roitblat, Search 2020: The Glossary.
  3. Herb Roitblat, Predictive Coding Glossary.
Print Friendly, PDF & Email