Software

The components are implemented following the JavaBeans naming conventions, and where applicable their actions are triggered by events which are passed from one component to the next. This component model is explained below. On this page you will also find information about the availability of the components, and a list of components which are on offer. If the right component for you is not in the list, there's always the option of getting custom software developed according to your requirements.

The software modules developed by Phrasys are built around a component model. They are implemented in 100% Java for maximum portability and follow the Java Beans conventions. Input and output are handled through events; these carry information that is passed from one component to the next, and each component processes this information before it is passed to the next.

For example, the Sentencer receives LineEvents from the LineReader component, converts these LineEvents into SentenceEvents and ParagraphEvents and passes them on to registered listeners. Later components in the processing chain use the WordEvent, which can be enriched by attribute value pairs, such as token=house and pos=noun.

Structural information can be handled via XML markup. XML tags are processed just like any other token, but they are obviously ignored when it comes to word class tagging or phrasal analysis. However, since they are part of the data stream the components can access them if required.

For building customised software a number of these components is chained together, and the custom part simply builds up on the final (or intermediate) output.

All components are available as shareware. For testing and evaluation purposes you can download restricted versions; these will mainly be identical with the registered version, apart from minor differences which are detailed in the description of the individual components.

Depending on how you are going to use the components, three different licences are available:

1. personal use only
2. non-profit distribution
3. commercial use

The list of components is constantly growing; at present it comprises

• LineReader, a basic component that generates LineEvents from an input file.
• Sentencer, to split (English) text into paragraphs and sentences.
• Tokeniser, to split text into individual words (or tokens).
• Tagger, to annotate text with word-class labels.
• Phraser, to identify (eg noun) phrases within the text.
• Stemmer, to reduce words to a canonical representation.
• Lemmatiser, to reduce words to their base form.

Note: during a transitional phase not all components are directly available for immediate downloading; if you require a component which is not yet `enabled' please send a mail to contact@phrasys.com or wait until it is available.

If you require NLP software components which are not listed above please contact us; depending on the specification it is possible for us to develop further software. Licensing and financial details are subject to negotiation.