SRL Editor Java program that lets you develop your own lightweight text mining systems using SRL
Download

SRL Editor Ranking & Summary

SRL Editor Tags

create edit extract text extractor text miner create expression

SRL Editor Description

Java program that lets you develop your own lightweight text mining systems using SRL The name of the SRL Editor is an acronym for the "Simple rule language" syntagm. SRL Editor is basically a program developed to perform fact extractions from a plain text and it was designed to operate as a two stage process. Firstly it extracts a set of predefined entity classes (phase named entity recognition) and then it uses these entities to fill in template slots by combining different contextual rules. As an example, the SRL Editor is currently being used in the BioCaster project at the National Institute of Informatics for identifying disease outbreaks and their locations from news reports.One of the key objectives of the SRL Editor has been to enable domain experts to develop their own lightweight text mining systems with minimal support from linguists and computer scientists. The SRL Editor program uses a powerful yet easy-to-understand regular expression language designed specifically for handling tokenized text streams. SRL Editor rules are often written using a simple combination of string literals and word classes (such as a list of country names) to define the matching context. If complexity is required the rules can easily incorporate constraints based on approximate matching, orthography, word distance etc. The SRL Editor, written in platform independent Java, supports user design of SRL rule sets using a graphic user interface (GUI) aiming for rapid development and test cycles.The SRL Editor aims to supply built in support for hand-crafted rule testing and revision, e.g. to find text segments where no rules are matching or to find rules which do not match any text. To enable fast creation and evaluation of rules the SRL editor allows the user to attach a corpus to the rule set, so that all matching contexts to a rule can be found rapidly and the user can instantly see if the rule is useful and correct. Future versions will aim to support improved rule revision mechanisms based on statistical metrics.Once SRL rule sets have been developed it is possible to run these in command-line mode on un-indexed text collections. Requirements: · Java

SRL Editor Related Software